Journal of Imaging

Research

11 pages, 1525 KiB

Open AccessArticle

Toward Closing the Loop in Image-to-Image Conversion in Radiotherapy: A Quality Control Tool to Predict Synthetic Computed Tomography Hounsfield Unit Accuracy

by Paolo Zaffino, Ciro Benito Raggio, Adrian Thummerer, Gabriel Guterres Marmitt, Johannes Albertus Langendijk, Anna Procopio, Carlo Cosentino, Joao Seco, Antje Christin Knopf, Stefan Both and Maria Francesca Spadea

J. Imaging 2024, 10(12), 316; https://doi.org/10.3390/jimaging10120316 - 10 Dec 2024

Viewed by 676

Abstract

In recent years, synthetic Computed Tomography (CT) images generated from Magnetic Resonance (MR) or Cone Beam Computed Tomography (CBCT) acquisitions have been shown to be comparable to real CT images in terms of dose computation for radiotherapy simulation. However, until now, there has [...] Read more.

In recent years, synthetic Computed Tomography (CT) images generated from Magnetic Resonance (MR) or Cone Beam Computed Tomography (CBCT) acquisitions have been shown to be comparable to real CT images in terms of dose computation for radiotherapy simulation. However, until now, there has been no independent strategy to assess the quality of each synthetic image in the absence of ground truth. In this work, we propose a Deep Learning (DL)-based framework to predict the accuracy of synthetic CT in terms of Mean Absolute Error (MAE) without the need for a ground truth (GT). The proposed algorithm generates a volumetric map as an output, informing clinicians of the predicted MAE slice-by-slice. A cascading multi-model architecture was used to deal with the complexity of the MAE prediction task. The workflow was trained and tested on two cohorts of head and neck cancer patients with different imaging modalities: 27 MR scans and 33 CBCT. The algorithm evaluation revealed an accurate HU prediction (a median absolute prediction deviation equal to 4 HU for CBCT-based synthetic CTs and 6 HU for MR-based synthetic CTs), with discrepancies that do not affect the clinical decisions made on the basis of the proposed estimation. The workflow exhibited no systematic error in MAE prediction. This work represents a proof of concept about the feasibility of synthetic CT evaluation in daily clinical practice, and it paves the way for future patient-specific quality assessment strategies. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Figure 1
Representation of the general MAE prediction pipeline. An axial sCT slice is given as input, and the associated MAE scalar for the image slice is predicted by using a DL pipeline. Full article ">Figure 2
A more detailed graphical representation of the MAE prediction pipeline. The final MAE prediction is obtained as a result of two DL steps: First a raw MAE interval classification is performed, followed by a more precise MAE estimation based on a regression algorithm. Full article ">Figure 3
Exemplary <math display="inline"><semantics> <mrow> <mi>s</mi> <mi>C</mi> <msub> <mi>T</mi> <mrow> <mi>C</mi> <mi>B</mi> <mi>C</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math> overlaid with its <math display="inline"><semantics> <mrow> <mi>p</mi> <mi>M</mi> <mi>A</mi> <msub> <mi>E</mi> <mrow> <mi>v</mi> <mi>o</mi> <mi>l</mi> <mi>u</mi> <mi>m</mi> <mi>e</mi> </mrow> </msub> </mrow> </semantics></math>. In addition to the 2D views (axial, sagittal, and coronal planes), the 3D representation is also shown. Full article ">Figure 4
Detailed workflow of MAE prediction. A single sCT axial slice is fed firstly into a DL model that classifies it as belonging to a specific MAE class. According to this prediction, the 2D image is then provided as input to a connected DL regression model, specifically trained to operate on a restricted range of MAE values. As a result, the MAE of a single sCT slice can be forecasted. In order to train the different models with a GT MAE, the ground truth CT is needed (dashed lines are needed only to train the models). Full article ">Figure 5
PD distributions for modality-specific and mixed pipelines. Results for <math display="inline"><semantics> <mrow> <mi>s</mi> <mi>C</mi> <msub> <mi>T</mi> <mrow> <mi>C</mi> <mi>B</mi> <mi>C</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <mi>s</mi> <mi>C</mi> <msub> <mi>T</mi> <mrow> <mi>M</mi> <mi>R</mi> </mrow> </msub> </mrow> </semantics></math> are reported, respectively, in the left and in the right panel. Full article ">Figure 6
APD distributions for modality-specific and mixed pipelines. Results for <math display="inline"><semantics> <mrow> <mi>s</mi> <mi>C</mi> <msub> <mi>T</mi> <mrow> <mi>C</mi> <mi>B</mi> <mi>C</mi> <mi>T</mi> </mrow> </msub> </mrow> </semantics></math> and <math display="inline"><semantics> <mrow> <mi>s</mi> <mi>C</mi> <msub> <mi>T</mi> <mrow> <mi>M</mi> <mi>R</mi> </mrow> </msub> </mrow> </semantics></math> are reported, respectively, in the left and in the right panel. Full article ">

21 pages, 2595 KiB

Open AccessArticle

Joint Image Processing with Learning-Driven Data Representation and Model Behavior for Non-Intrusive Anemia Diagnosis in Pediatric Patients

by Tarek Berghout

J. Imaging 2024, 10(10), 245; https://doi.org/10.3390/jimaging10100245 - 2 Oct 2024

Cited by 1 | Viewed by 1376

Abstract

Anemia diagnosis is crucial for pediatric patients due to its impact on growth and development. Traditional methods, like blood tests, are effective but pose challenges, such as discomfort, infection risk, and frequent monitoring difficulties, underscoring the need for non-intrusive diagnostic methods. In light [...] Read more.

Anemia diagnosis is crucial for pediatric patients due to its impact on growth and development. Traditional methods, like blood tests, are effective but pose challenges, such as discomfort, infection risk, and frequent monitoring difficulties, underscoring the need for non-intrusive diagnostic methods. In light of this, this study proposes a novel method that combines image processing with learning-driven data representation and model behavior for non-intrusive anemia diagnosis in pediatric patients. The contributions of this study are threefold. First, it uses an image-processing pipeline to extract 181 features from 13 categories, with a feature-selection process identifying the most crucial data for learning. Second, a deep multilayered network based on long short-term memory (LSTM) is utilized to train a model for classifying images into anemic and non-anemic cases, where hyperparameters are optimized using Bayesian approaches. Third, the trained LSTM model is integrated as a layer into a learning model developed based on recurrent expansion rules, forming a part of a new deep network called a recurrent expansion network (RexNet). RexNet is designed to learn data representations akin to traditional deep-learning methods while also understanding the interaction between dependent and independent variables. The proposed approach is applied to three public datasets, namely conjunctival eye images, palmar images, and fingernail images of children aged up to 6 years. RexNet achieves an overall evaluation of 99.83 ± 0.02% across all classification metrics, demonstrating significant improvements in diagnostic results and generalization compared to LSTM networks and existing methods. This highlights RexNet’s potential as a promising alternative to traditional blood-based methods for non-intrusive anemia diagnosis. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Figure 1
Example photographs of collected data and regions of interest. (a–c) Eye conjunctiva, hand palm, and fingernail images of non-anemic individuals; (d–f) eye conjunctiva, hand palm, and fingernail images of anemic individuals, with green zones highlighting the regions of interest in this study. Adapted from [<a href="#B20-jimaging-10-00245" class="html-bibr">20</a>]: WILEY 2023, under open access license. The figure has been modified for better clarity, including realignment, recoloring of regions of interest, and denoising. Full article ">Figure 2
Flowchart of the proposed methodology for image processing. Full article ">Figure 3
Class proportions of image datasets: (a) palmar images; (b) eye conjunctiva images; and (c) fingernail images. Full article ">Figure 4
Selected feature categories and their proportions based on the feature-extraction process for each dataset. (a,b) Proportions of selected feature categories for the palm dataset; (c,d) proportions of selected feature categories for the eye conjunctiva dataset; and (e,f) proportions of selected feature categories for the Fingernails dataset. Full article ">Figure 5
Dataset class scatters after image processing. (a) Palmar images; (b) eye conjunctiva images; and (c) fingernail images. Full article ">Figure 6
Flow diagram of proposed RexNet. Full article ">Figure 7
Training behavior of the studied approaches. (a) Loss function of RexNet and LSTM for the fingernails dataset; (b) loss function of RexNet and LSTM for the eye conjunctiva dataset; and (c) loss function of RexNet and LSTM for the palmar images dataset. Full article ">Figure 8
ROC curves behavior of the studied approaches. (a) ROC curves of RexNet and LSTM for the fingernails dataset; (b) ROC curves of RexNet and LSTM for the eye conjunctiva dataset; (c) ROC curves of RexNet and LSTM for the palmar images dataset; and (d) zoomed-in subplot for dataset 2—palm dataset. Full article ">Figure 9
Confusion matrices for LSTM and RexNet models across different datasets. (a–c) LSTM results on fingernails, palm, and conjunctival eye datasets; (d–f) RexNet results for fingernails, palm, and conjunctival eye datasets. Full article ">

10 pages, 5992 KiB

Open AccessArticle

Comparison of Visual and Quantra Software Mammographic Density Assessment According to BI-RADS^® in 2D and 3D Images

by Francesca Morciano, Cristina Marcazzan, Rossella Rella, Oscar Tommasini, Marco Conti, Paolo Belli, Andrea Spagnolo, Andrea Quaglia, Stefano Tambalo, Andreea Georgiana Trisca, Claudia Rossati, Francesca Fornasa and Giovanna Romanucci

J. Imaging 2024, 10(9), 238; https://doi.org/10.3390/jimaging10090238 - 23 Sep 2024

Viewed by 815

Abstract

Mammographic density (MD) assessment is subject to inter- and intra-observer variability. An automated method, such as Quantra software, could be a useful tool for an objective and reproducible MD assessment. Our purpose was to evaluate the performance of Quantra software in assessing MD, [...] Read more.

Mammographic density (MD) assessment is subject to inter- and intra-observer variability. An automated method, such as Quantra software, could be a useful tool for an objective and reproducible MD assessment. Our purpose was to evaluate the performance of Quantra software in assessing MD, according to BI-RADS^® Atlas Fifth Edition recommendations, verifying the degree of agreement with the gold standard, given by the consensus of two breast radiologists. A total of 5009 screening examinations were evaluated by two radiologists and analysed by Quantra software to assess MD. The agreement between the three assigned values was expressed as intraclass correlation coefficients (ICCs). The agreement between the software and the two readers (R1 and R2) was moderate with ICC values of 0.725 and 0.713, respectively. A better agreement was demonstrated between the software’s assessment and the average score of the values assigned by the two radiologists, with an index of 0.793, which reflects a good correlation. Quantra software appears a promising tool in supporting radiologists in the MD assessment and could be part of a personalised screening protocol soon. However, some fine-tuning is needed to improve its accuracy, reduce its tendency to overestimate, and ensure it excludes high-density structures from its assessment. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Figure 1
Raw images for MD assessment by Quantra Software. Full article ">Figure 2
RCC, LCC, RMLO, and LMLO 2D-synthetic mammograms assigned to category A, “almost entirely fat”, by the Quantra 2.2.3 software according to BI-RADS lexicon. Full article ">Figure 3
RCC, LCC, RMLO, and LMLO 2D-synthetic mammograms assigned to category B, “scattered fibroglandular”, by the Quantra 2.2.3 software according to BI-RADS lexicon. Full article ">Figure 4
RCC, LCC, RMLO, and LMLO 2D-synthetic mammograms assigned to category C, “heterogeneously dense”, by the Quantra 2.2.3 software according to BI-RADS lexicon. Full article ">Figure 5
RCC, LCC, RMLO, and LMLO 2D-synthetic mammograms assigned to category D, “extremely dense”, by the Quantra 2.2.3 software according to BI-RADS lexicon. Full article ">Figure 6
Density categories assigned by Quantra Software according to BI-RADS lexicon, fifth edition. Full article ">Figure 7
RCC, LCC, RMLO, and LMLO 2D-synthetic mammograms of the same patient. The visual assessment made by radiologists for this exam was “B”. Quantra software assigned the highest category, “D”, due to the presence of breast implants, resulting in an incorrect assessment. Full article ">Figure 8
RCC, LCC, RMLO, and LMLO 2D-synthetic mammograms of the same patient. The visual assessment made by radiologists for this exam was “B”. Quantra software assigned the highest category, “D”, due to the presence of a loop recorder and some macrocalcifications in the left breast, resulting in an incorrect assessment. Full article ">Chart 1
Graphic illustration of the data reported in <a href="#jimaging-10-00238-t001" class="html-table">Table 1</a> on the frequency of breast density categories assigned by each reader (R1 and R2), the mean of the two readers (R1-R2), and the Quantra software (R3). Full article ">

23 pages, 5832 KiB

Open AccessArticle

Enhancing Deep Learning Model Explainability in Brain Tumor Datasets Using Post-Heuristic Approaches

by Konstantinos Pasvantis and Eftychios Protopapadakis

J. Imaging 2024, 10(9), 232; https://doi.org/10.3390/jimaging10090232 - 18 Sep 2024

Viewed by 1264

Abstract

The application of deep learning models in medical diagnosis has showcased considerable efficacy in recent years. Nevertheless, a notable limitation involves the inherent lack of explainability during decision-making processes. This study addresses such a constraint by enhancing the interpretability robustness. The primary focus [...] Read more.

The application of deep learning models in medical diagnosis has showcased considerable efficacy in recent years. Nevertheless, a notable limitation involves the inherent lack of explainability during decision-making processes. This study addresses such a constraint by enhancing the interpretability robustness. The primary focus is directed towards refining the explanations generated by the LIME Library and LIME image explainer. This is achieved through post-processing mechanisms based on scenario-specific rules. Multiple experiments have been conducted using publicly accessible datasets related to brain tumor detection. Our proposed post-heuristic approach demonstrates significant advancements, yielding more robust and concrete results in the context of medical diagnosis. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Graphical abstract

13 pages, 1308 KiB

Open AccessArticle

Decoding Breast Cancer: Using Radiomics to Non-Invasively Unveil Molecular Subtypes Directly from Mammographic Images

by Manon A. G. Bakker, Maria de Lurdes Ovalho, Nuno Matela and Ana M. Mota

J. Imaging 2024, 10(9), 218; https://doi.org/10.3390/jimaging10090218 - 4 Sep 2024

Viewed by 1435

Abstract

Breast cancer is the most commonly diagnosed cancer worldwide. The therapy used and its success depend highly on the histology of the tumor. This study aimed to explore the potential of predicting the molecular subtype of breast cancer using radiomic features extracted from [...] Read more.

Breast cancer is the most commonly diagnosed cancer worldwide. The therapy used and its success depend highly on the histology of the tumor. This study aimed to explore the potential of predicting the molecular subtype of breast cancer using radiomic features extracted from screening digital mammography (DM) images. A retrospective study was performed using the OPTIMAM Mammography Image Database (OMI-DB). Four binary classification tasks were performed: luminal A vs. non-luminal A, luminal B vs. non-luminal B, TNBC vs. non-TNBC, and HER2 vs. non-HER2. Feature selection was carried out by Pearson correlation and LASSO. The support vector machine (SVM) and naive Bayes (NB) ML classifiers were used, and their performance was evaluated with the accuracy and the area under the receiver operating characteristic curve (AUC). A total of 186 patients were included in the study: 58 luminal A, 35 luminal B, 52 TNBC, and 41 HER2. The SVM classifier resulted in AUCs during testing of 0.855 for luminal A, 0.812 for luminal B, 0.789 for TNBC, and 0.755 for HER2, respectively. The NB classifier showed AUCs during testing of 0.714 for luminal A, 0.746 for luminal B, 0.593 for TNBC, and 0.714 for HER2. The SVM classifier outperformed NB with statistical significance for luminal A (p = 0.0268) and TNBC (p = 0.0073). Our study showed the potential of radiomics for non-invasive breast cancer subtype classification. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Figure 1
The in- and exclusion criteria flowchart used during this study. Full article ">Figure 2
The tumor segmentation process. Starting with normalization of the original DM image, where breast lesions (red) classified as ‘calcification’ underwent image enhancement. Breast lesions classified as ‘mass’ underwent segmentation using a region-growing algorithm. The segmentations were finalized with the use of the image segmenter tool from MATLAB to obtain the final tumor segmentation. Full article ">Figure 3
An example of image enhancement for a calcification region where (a) is the original DM image and (b) is the enhanced image, making the calcification more pronounced. Full article ">Figure 4
Examples of breast tumor segmentations for (a) luminal A, (b) luminal B, (c) TNBC, and (d) HER2. Full article ">Figure 5
The selected radiomic features for (a) luminal A vs. non-luminal A, (b) luminal B vs. non-luminal B, (c) TNBC vs. non-TNBC, and (d) HER2 vs. non-HER2 classification tasks. Full article ">Figure 6
The ROC curves of the SVM (blue) and NB (yellow) classifiers for (a) luminal A vs. non-luminal A, (b) luminal B vs. non-luminal B, (c) TNBC vs. non-TNBC, and (d) HER2 vs. non-HER2. Full article ">

29 pages, 4861 KiB

Open AccessArticle

A New Approach for Effective Retrieval of Medical Images: A Step towards Computer-Assisted Diagnosis

by Suchita Sharma and Ashutosh Aggarwal

J. Imaging 2024, 10(9), 210; https://doi.org/10.3390/jimaging10090210 - 26 Aug 2024

Viewed by 868

Abstract

The biomedical imaging field has grown enormously in the past decade. In the era of digitization, the demand for computer-assisted diagnosis is increasing day by day. The COVID-19 pandemic further emphasized how retrieving meaningful information from medical repositories can aid in improving the [...] Read more.

The biomedical imaging field has grown enormously in the past decade. In the era of digitization, the demand for computer-assisted diagnosis is increasing day by day. The COVID-19 pandemic further emphasized how retrieving meaningful information from medical repositories can aid in improving the quality of patient’s diagnosis. Therefore, content-based retrieval of medical images has a very prominent role in fulfilling our ultimate goal of developing automated computer-assisted diagnosis systems. Therefore, this paper presents a content-based medical image retrieval system that extracts multi-resolution, noise-resistant, rotation-invariant texture features in the form of a novel pattern descriptor, i.e.,

M s N r R i T x P

, from medical images. In the proposed approach, the input medical image is initially decomposed into three neutrosophic images on its transformation into the neutrosophic domain. Afterwards, three distinct pattern descriptors, i.e.,

M s T r P

,

N r T x P

, and

R i T x P

, are derived at multiple scales from the three neutrosophic images. The proposed

M s N r R i T x P

pattern descriptor is obtained by scale-wise concatenation of the joint histograms of

M s T r P \times R i T x P

and

N r T x P \times R i T x P

. To demonstrate the efficacy of the proposed system, medical images of different modalities, i.e., CT and MRI, from four test datasets are considered in our experimental setup. The retrieval performance of the proposed approach is exhaustively compared with several existing, recent, and state-of-the-art local binary pattern-based variants. The retrieval rates obtained by the proposed approach for the noise-free and noisy variants of the test datasets are observed to be substantially higher than the compared ones. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

27 pages, 14394 KiB

Open AccessArticle

Celiac Disease Deep Learning Image Classification Using Convolutional Neural Networks

by Joaquim Carreras

J. Imaging 2024, 10(8), 200; https://doi.org/10.3390/jimaging10080200 - 16 Aug 2024

Cited by 1 | Viewed by 1659

Abstract

Celiac disease (CD) is a gluten-sensitive immune-mediated enteropathy. This proof-of-concept study used a convolutional neural network (CNN) to classify hematoxylin and eosin (H&E) CD histological images, normal small intestine control, and non-specified duodenal inflammation (7294, 11,642, and 5966 images, respectively). The trained network [...] Read more.

Celiac disease (CD) is a gluten-sensitive immune-mediated enteropathy. This proof-of-concept study used a convolutional neural network (CNN) to classify hematoxylin and eosin (H&E) CD histological images, normal small intestine control, and non-specified duodenal inflammation (7294, 11,642, and 5966 images, respectively). The trained network classified CD with high performance (accuracy 99.7%, precision 99.6%, recall 99.3%, F1-score 99.5%, and specificity 99.8%). Interestingly, when the same network (already trained for the 3 class images), analyzed duodenal adenocarcinoma (3723 images), the new images were classified as duodenal inflammation in 63.65%, small intestine control in 34.73%, and CD in 1.61% of the cases; and when the network was retrained using the 4 histological subtypes, the performance was above 99% for CD and 97% for adenocarcinoma. Finally, the model added 13,043 images of Crohn’s disease to include other inflammatory bowel diseases; a comparison between different CNN architectures was performed, and the gradient-weighted class activation mapping (Grad-CAM) technique was used to understand why the deep learning network made its classification decisions. In conclusion, the CNN-based deep neural system classified 5 diagnoses with high performance. Narrow artificial intelligence (AI) is designed to perform tasks that typically require human intelligence, but it operates within limited constraints and is task-specific. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Graphical abstract

24 pages, 410 KiB

Open AccessArticle

Gastric Cancer Image Classification: A Comparative Analysis and Feature Fusion Strategies

by Andrea Loddo, Marco Usai and Cecilia Di Ruberto

J. Imaging 2024, 10(8), 195; https://doi.org/10.3390/jimaging10080195 - 10 Aug 2024

Cited by 1 | Viewed by 1424

Abstract

Gastric cancer is the fifth most common and fourth deadliest cancer worldwide, with a bleak 5-year survival rate of about 20%. Despite significant research into its pathobiology, prognostic predictability remains insufficient due to pathologists’ heavy workloads and the potential for diagnostic errors. Consequently, [...] Read more.

Gastric cancer is the fifth most common and fourth deadliest cancer worldwide, with a bleak 5-year survival rate of about 20%. Despite significant research into its pathobiology, prognostic predictability remains insufficient due to pathologists’ heavy workloads and the potential for diagnostic errors. Consequently, there is a pressing need for automated and precise histopathological diagnostic tools. This study leverages Machine Learning and Deep Learning techniques to classify histopathological images into healthy and cancerous categories. By utilizing both handcrafted and deep features and shallow learning classifiers on the GasHisSDB dataset, we conduct a comparative analysis to identify the most effective combinations of features and classifiers for differentiating normal from abnormal histopathological images without employing fine-tuning strategies. Our methodology achieves an accuracy of 95% with the SVM classifier, underscoring the effectiveness of feature fusion strategies. Additionally, cross-magnification experiments produced promising results with accuracies close to 80% and 90% when testing the models on unseen testing images with different resolutions. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

31 pages, 5788 KiB

Open AccessArticle

Automated Lung Cancer Diagnosis Applying Butterworth Filtering, Bi-Level Feature Extraction, and Sparce Convolutional Neural Network to Luna 16 CT Images

by Nasr Y. Gharaibeh, Roberto De Fazio, Bassam Al-Naami, Abdel-Razzak Al-Hinnawi and Paolo Visconti

J. Imaging 2024, 10(7), 168; https://doi.org/10.3390/jimaging10070168 - 15 Jul 2024

Viewed by 1718

Abstract

Accurate prognosis and diagnosis are crucial for selecting and planning lung cancer treatments. As a result of the rapid development of medical imaging technology, the use of computed tomography (CT) scans in pathology is becoming standard practice. An intricate interplay of requirements and [...] Read more.

Accurate prognosis and diagnosis are crucial for selecting and planning lung cancer treatments. As a result of the rapid development of medical imaging technology, the use of computed tomography (CT) scans in pathology is becoming standard practice. An intricate interplay of requirements and obstacles characterizes computer-assisted diagnosis, which relies on the precise and effective analysis of pathology images. In recent years, pathology image analysis tasks such as tumor region identification, prognosis prediction, tumor microenvironment characterization, and metastasis detection have witnessed the considerable potential of artificial intelligence, especially deep learning techniques. In this context, an artificial intelligence (AI)-based methodology for lung cancer diagnosis is proposed in this research work. As a first processing step, filtering using the Butterworth smooth filter algorithm was applied to the input images from the LUNA 16 lung cancer dataset to remove noise without significantly degrading the image quality. Next, we performed the bi-level feature selection step using the Chaotic Crow Search Algorithm and Random Forest (CCSA-RF) approach to select features such as diameter, margin, spiculation, lobulation, subtlety, and malignancy. Next, the Feature Extraction step was performed using the Multi-space Image Reconstruction (MIR) method with Grey Level Co-occurrence Matrix (GLCM). Next, the Lung Tumor Severity Classification (LTSC) was implemented by using the Sparse Convolutional Neural Network (SCNN) approach with a Probabilistic Neural Network (PNN). The developed method can detect benign, normal, and malignant lung cancer images using the PNN algorithm, which reduces complexity and efficiently provides classification results. Performance parameters, namely accuracy, precision, F-score, sensitivity, and specificity, were determined to evaluate the effectiveness of the implemented hybrid method and compare it with other solutions already present in the literature. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

21 pages, 1918 KiB

Open AccessArticle

Residual-Based Multi-Stage Deep Learning Framework for Computer-Aided Alzheimer’s Disease Detection

by Najmul Hassan, Abu Saleh Musa Miah and Jungpil Shin

J. Imaging 2024, 10(6), 141; https://doi.org/10.3390/jimaging10060141 - 11 Jun 2024

Viewed by 1894

Abstract

Alzheimer’s Disease (AD) poses a significant health risk globally, particularly among the elderly population. Recent studies underscore its prevalence, with over 50% of elderly Japanese facing a lifetime risk of dementia, primarily attributed to AD. As the most prevalent form of dementia, AD [...] Read more.

Alzheimer’s Disease (AD) poses a significant health risk globally, particularly among the elderly population. Recent studies underscore its prevalence, with over 50% of elderly Japanese facing a lifetime risk of dementia, primarily attributed to AD. As the most prevalent form of dementia, AD gradually erodes brain cells, leading to severe neurological decline. In this scenario, it is important to develop an automatic AD-detection system, and many researchers have been working to develop an AD-detection system by taking advantage of the advancement of deep learning (DL) techniques, which have shown promising results in various domains, including medical image analysis. However, existing approaches for AD detection often suffer from limited performance due to the complexities associated with training hierarchical convolutional neural networks (CNNs). In this paper, we introduce a novel multi-stage deep neural network architecture based on residual functions to address the limitations of existing AD-detection approaches. Inspired by the success of residual networks (ResNets) in image-classification tasks, our proposed system comprises five stages, each explicitly formulated to enhance feature effectiveness while maintaining model depth. Following feature extraction, a deep learning-based feature-selection module is applied to mitigate overfitting, incorporating batch normalization, dropout and fully connected layers. Subsequently, machine learning (ML)-based classification algorithms, including Support Vector Machines (SVM), Random Forest (RF) and SoftMax, are employed for classification tasks. Comprehensive evaluations conducted on three benchmark datasets, namely ADNI1: Complete 1Yr 1.5T, MIRAID and OASIS Kaggle, demonstrate the efficacy of our proposed model. Impressively, our model achieves accuracy rates of 99.47%, 99.10% and 99.70% for ADNI1: Complete 1Yr 1.5T, MIRAID and OASIS datasets, respectively, outperforming existing systems in binary class problems. Our proposed model represents a significant advancement in the AD-analysis domain. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

21 pages, 12872 KiB

Open AccessArticle

Optimizing Vision Transformers for Histopathology: Pretraining and Normalization in Breast Cancer Classification

by Giulia Lucrezia Baroni, Laura Rasotto, Kevin Roitero, Angelica Tulisso, Carla Di Loreto and Vincenzo Della Mea

J. Imaging 2024, 10(5), 108; https://doi.org/10.3390/jimaging10050108 - 30 Apr 2024

Cited by 3 | Viewed by 2400

Abstract

This paper introduces a self-attention Vision Transformer model specifically developed for classifying breast cancer in histology images. We examine various training strategies and configurations, including pretraining, dimension resizing, data augmentation and color normalization strategies, patch overlap, and patch size configurations, in order to [...] Read more.

This paper introduces a self-attention Vision Transformer model specifically developed for classifying breast cancer in histology images. We examine various training strategies and configurations, including pretraining, dimension resizing, data augmentation and color normalization strategies, patch overlap, and patch size configurations, in order to evaluate their impact on the effectiveness of the histology image classification. Additionally, we provide evidence for the increase in effectiveness gathered through geometric and color data augmentation techniques. We primarily utilize the BACH dataset to train and validate our methods and models, but we also test them on two additional datasets, BRACS and AIDPATH, to verify their generalization capabilities. Our model, developed from a transformer pretrained on ImageNet, achieves an accuracy rate of 0.91 on the BACH dataset, 0.74 on the BRACS dataset, and 0.92 on the AIDPATH dataset. Using a model based on the prostate small and prostate medium HistoEncoder models, we achieve accuracy rates of 0.89 and 0.86, respectively. Our results suggest that pretraining on large-scale general datasets like ImageNet is advantageous. We also show the potential benefits of using domain-specific pretraining datasets, such as extensive histopathological image collections as in HistoEncoder, though not yet with clear advantages. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

14 pages, 5416 KiB

Open AccessArticle

Attention-Enhanced Unpaired xAI-GANs for Transformation of Histological Stain Images

by Tibor Sloboda, Lukáš Hudec, Matej Halinkovič and Wanda Benesova

J. Imaging 2024, 10(2), 32; https://doi.org/10.3390/jimaging10020032 - 25 Jan 2024

Cited by 1 | Viewed by 2167

Abstract

Histological staining is the primary method for confirming cancer diagnoses, but certain types, such as p63 staining, can be expensive and potentially damaging to tissues. In our research, we innovate by generating p63-stained images from H&E-stained slides for metaplastic breast cancer. This is [...] Read more.

Histological staining is the primary method for confirming cancer diagnoses, but certain types, such as p63 staining, can be expensive and potentially damaging to tissues. In our research, we innovate by generating p63-stained images from H&E-stained slides for metaplastic breast cancer. This is a crucial development, considering the high costs and tissue risks associated with direct p63 staining. Our approach employs an advanced CycleGAN architecture, xAI-CycleGAN, enhanced with context-based loss to maintain structural integrity. The inclusion of convolutional attention in our model distinguishes between structural and color details more effectively, thus significantly enhancing the visual quality of the results. This approach shows a marked improvement over the base xAI-CycleGAN and standard CycleGAN models, offering the benefits of a more compact network and faster training even with the inclusion of attention. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Figure 1
Demonstration of significant differences in tissue in paired and aligned p63 stained tissue (left) compared with its H&E counterpart (right) [<a href="#B9-jimaging-10-00032" class="html-bibr">9</a>]. Full article ">Figure 2
Demonstration of the context loss computation in one direction. <math display="inline"><semantics> <msup> <mi>G</mi> <mrow> <mi>A</mi> <mi>B</mi> </mrow> </msup> </semantics></math> represents the H&E to p63 conversion and <math display="inline"><semantics> <msup> <mi>G</mi> <mrow> <mi>B</mi> <mi>A</mi> </mrow> </msup> </semantics></math> represents the p63 to H&E conversion. <math display="inline"><semantics> <msup> <mi>B</mi> <mo>′</mo> </msup> </semantics></math> is the converted (fake) p63 image and <math display="inline"><semantics> <msup> <mi>A</mi> <mi>c</mi> </msup> </semantics></math> is the cycled H&E image. H represents Huber Loss. Full article ">Figure 3
Results show that the FID of the enhanced xAI-CycleGAN is significantly lower than that of the original, demonstrating significant improvement over the previous method. Full article ">Figure 4
Comparing conversion from p63 to H&E for both original and improved xAI-CycleGAN at the same point in training, demonstrating that the issue with corruption/artifacts in conversion is solved. (a) Demonstration of the corrupted conversion present in original xAI-CycleGAN due to explainability-driven training. (b) Demonstration of a clean proper conversion with our enhanced model, using the same test image, at the same point during training (the model has seen the same amount of training samples). Full article ">Figure 5
A demonstration of successful editing capabilities. (A) contains image converted from H&E to p63 without any modifications applied. (B) contains an image with modifications applied to best match the real image. (C) contains the unmodified original p63 image of the same region. The same base image as in our previous work [<a href="#B9-jimaging-10-00032" class="html-bibr">9</a>] has been used. Full article ">Figure 6
Demonstration of conversion by our model from H&E to p63 where no myoepithelial cells are present. (a) Raw unedited H&E that has no myoepithelial cells present. (b) Edited p63 image after correct transformation to include brown coloring by p63 for myoepithelial cells, which is not meant to be present in the image. For this image, we used <math display="inline"><semantics> <mrow> <mi>r</mi> <mo>=</mo> <mn>3</mn> </mrow> </semantics></math> with <math display="inline"><semantics> <mrow> <mi>α</mi> <mo>=</mo> <mo>[</mo> <mo>−</mo> <mn>2</mn> <mo>,</mo> <mo>−</mo> <mn>1.6</mn> <mo>,</mo> <mn>0.2</mn> <mo>]</mo> </mrow> </semantics></math> for each vector, respectively. Full article ">Figure 7
A grid of edited images with varying r from 1 to 5, and <math display="inline"><semantics> <mi>α</mi> </semantics></math> values for columns ranging from −5.0 to 5.0 with a step size of 2.5 (applied to all matrices V defined in Equation (<a href="#FD8-jimaging-10-00032" class="html-disp-formula">8</a>)). Full article ">

Review

Jump to: Research, Other

54 pages, 5089 KiB

Open AccessReview

The Neural Frontier of Future Medical Imaging: A Review of Deep Learning for Brain Tumor Detection

by Tarek Berghout

J. Imaging 2025, 11(1), 2; https://doi.org/10.3390/jimaging11010002 - 24 Dec 2024

Viewed by 433

Abstract

Brain tumor detection is crucial in medical research due to high mortality rates and treatment challenges. Early and accurate diagnosis is vital for improving patient outcomes, however, traditional methods, such as manual Magnetic Resonance Imaging (MRI) analysis, are often time-consuming and error-prone. The [...] Read more.

Brain tumor detection is crucial in medical research due to high mortality rates and treatment challenges. Early and accurate diagnosis is vital for improving patient outcomes, however, traditional methods, such as manual Magnetic Resonance Imaging (MRI) analysis, are often time-consuming and error-prone. The rise of deep learning has led to advanced models for automated brain tumor feature extraction, segmentation, and classification. Despite these advancements, comprehensive reviews synthesizing recent findings remain scarce. By analyzing over 100 research papers over past half-decade (2019–2024), this review fills that gap, exploring the latest methods and paradigms, summarizing key concepts, challenges, datasets, and offering insights into future directions for brain tumor detection using deep learning. This review also incorporates an analysis of previous reviews and targets three main aspects: feature extraction, segmentation, and classification. The results revealed that research primarily focuses on Convolutional Neural Networks (CNNs) and their variants, with a strong emphasis on transfer learning using pre-trained models. Other methods, such as Generative Adversarial Networks (GANs) and Autoencoders, are used for feature extraction, while Recurrent Neural Networks (RNNs) are employed for time-sequence modeling. Some models integrate with Internet of Things (IoT) frameworks or federated learning for real-time diagnostics and privacy, often paired with optimization algorithms. However, the adoption of eXplainable AI (XAI) remains limited, despite its importance in building trust in medical diagnostics. Finally, this review outlines future opportunities, focusing on image quality, underexplored deep learning techniques, expanding datasets, and exploring deeper learning representations and model behavior such as recurrent expansion to advance medical imaging diagnostics. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

35 pages, 7878 KiB

Open AccessReview

Advances in Real-Time 3D Reconstruction for Medical Endoscopy

by Alexander Richter, Till Steinmann, Jean-Claude Rosenthal and Stefan J. Rupitsch

J. Imaging 2024, 10(5), 120; https://doi.org/10.3390/jimaging10050120 - 14 May 2024

Viewed by 3343

Abstract

This contribution is intended to provide researchers with a comprehensive overview of the current state-of-the-art concerning real-time 3D reconstruction methods suitable for medical endoscopy. Over the past decade, there have been various technological advancements in computational power and an increased research effort in [...] Read more.

This contribution is intended to provide researchers with a comprehensive overview of the current state-of-the-art concerning real-time 3D reconstruction methods suitable for medical endoscopy. Over the past decade, there have been various technological advancements in computational power and an increased research effort in many computer vision fields such as autonomous driving, robotics, and unmanned aerial vehicles. Some of these advancements can also be adapted to the field of medical endoscopy while coping with challenges such as featureless surfaces, varying lighting conditions, and deformable structures. To provide a comprehensive overview, a logical division of monocular, binocular, trinocular, and multiocular methods is performed and also active and passive methods are distinguished. Within these categories, we consider both flexible and non-flexible endoscopes to cover the state-of-the-art as fully as possible. The relevant error metrics to compare the publications presented here are discussed, and the choice of when to choose a GPU rather than an FPGA for camera-based 3D reconstruction is debated. We elaborate on the good practice of using datasets and provide a direct comparison of the presented work. It is important to note that in addition to medical publications, publications evaluated on the KITTI and Middlebury datasets are also considered to include related methods that may be suited for medical 3D reconstruction. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Other

Jump to: Research, Review

10 pages, 1746 KiB

Open AccessTechnical Note

MOTH: Memory-Efficient On-the-Fly Tiling of Histological Image Annotations Using QuPath

by Thomas Kauer, Jannik Sehring, Kai Schmid, Marek Bartkuhn, Benedikt Wiebach, Slaven Crnkovic, Grazyna Kwapiszewska, Till Acker and Daniel Amsel

J. Imaging 2024, 10(11), 292; https://doi.org/10.3390/jimaging10110292 - 15 Nov 2024

Viewed by 775

Abstract

The emerging usage of digitalized histopathological images is leading to a novel possibility for data analysis. With the help of artificial intelligence algorithms, it is now possible to detect certain structures and morphological features on whole slide images automatically. This enables algorithms to [...] Read more.

The emerging usage of digitalized histopathological images is leading to a novel possibility for data analysis. With the help of artificial intelligence algorithms, it is now possible to detect certain structures and morphological features on whole slide images automatically. This enables algorithms to count, measure, or evaluate those areas when trained properly. To achieve suitable training, datasets must be annotated and curated by users in programs like QuPath. The extraction of this data for artificial intelligence algorithms is still rather tedious and needs to be saved on a local hard drive. We developed a toolkit for integration into existing pipelines and tools, like U-net, for the on-the-fly extraction of annotation tiles from existing QuPath projects. The tiles can be directly used as input for artificial intelligence algorithms, and the results are directly transferred back to QuPath for visual inspection. With the toolkit, we created a convenient way to incorporate QuPath into existing AI workflows. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Figure 1
MOTH overview. MOTH is a suite of tools that facilitates the import and export of annotations and images from and into QuPath. The system is capable of establishing a connection to local AI-based algorithms. Full article ">Figure 2
(A,B) IoU and HD of exported shapes rendered with MOTH and Groovy in the artificial dataset. (C,D) IoU and HD of exported shapes rendered with MOTH and Groovy in the mitosis dataset. Groovy results are marked in orange and MOTH results are marked in green. Diamonds represent outliers. Full article ">Figure 3
MOTH export of small shapes with pixel offsets. The figure shows the export of small ground truth shapes. The ground truth shapes are drawn as orange lines and the center of the shape is marked by an orange dot. Black areas are pixels set in the MOTH export. A high overlap with the ground truth shapes can be observed. Full article ">Figure 4
Groovy export of small shapes with pixel offsets. The figure shows the export of small ground truth shapes. In comparison to the previous figure, a lower overlap between the ground truth and the export is visible. Full article ">Figure 5
Real world example using MOTH. The proposals are generated via QuPath and extracted from the project via MOTH. The proposals are evaluated and improved via custom methods and loaded back into QuPath for visual inspection using MOTH. Full article ">

22 pages, 604 KiB

Open AccessSystematic Review

The Accuracy of Three-Dimensional Soft Tissue Simulation in Orthognathic Surgery—A Systematic Review

by Anna Olejnik, Laurence Verstraete, Tomas-Marijn Croonenborghs, Constantinus Politis and Gwen R. J. Swennen

J. Imaging 2024, 10(5), 119; https://doi.org/10.3390/jimaging10050119 - 14 May 2024

Cited by 1 | Viewed by 1486

Abstract

Three-dimensional soft tissue simulation has become a popular tool in the process of virtual orthognathic surgery planning and patient–surgeon communication. To apply 3D soft tissue simulation software in routine clinical practice, both qualitative and quantitative validation of its accuracy are required. The objective [...] Read more.

Three-dimensional soft tissue simulation has become a popular tool in the process of virtual orthognathic surgery planning and patient–surgeon communication. To apply 3D soft tissue simulation software in routine clinical practice, both qualitative and quantitative validation of its accuracy are required. The objective of this study was to systematically review the literature on the accuracy of 3D soft tissue simulation in orthognathic surgery. The Web of Science, PubMed, Cochrane, and Embase databases were consulted for the literature search. The systematic review (SR) was conducted according to the PRISMA statement, and 40 articles fulfilled the inclusion and exclusion criteria. The Quadas-2 tool was used for the risk of bias assessment for selected studies. A mean error varying from 0.27 mm to 2.9 mm for 3D soft tissue simulations for the whole face was reported. In the studies evaluating 3D soft tissue simulation accuracy after a Le Fort I osteotomy only, the upper lip and paranasal regions were reported to have the largest error, while after an isolated bilateral sagittal split osteotomy, the largest error was reported for the lower lip and chin regions. In the studies evaluating simulation after bimaxillary osteotomy with or without genioplasty, the highest inaccuracy was reported at the level of the lips, predominantly the lower lip, chin, and, sometimes, the paranasal regions. Due to the variability in the study designs and analysis methods, a direct comparison was not possible. Therefore, based on the results of this SR, guidelines to systematize the workflow for evaluating the accuracy of 3D soft tissue simulations in orthognathic surgery in future studies are proposed. Full article

(This article belongs to the Special Issue Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Advances in Biomedical Image Processing and Artificial Intelligence for Computer-Aided Diagnosis in Medicine

Share This Special Issue

Special Issue Editors

Special Issue Information

Keywords

Benefits of Publishing in a Special Issue

Published Papers (16 papers)

Research

Review

Other

Further Information

Guidelines

MDPI Initiatives

Follow MDPI