Feasibility study of ResNet-50 in the distinction of intraoral neural tumors using histopathological images.

BACKGROUND: Neural tumors are difficult to distinguish based solely on cellularity and often require immunohistochemical staining to aid in identifying the cell lineage. This article investigates the potential of a Convolutional Neural Network for the histopathological classification of the three most prevalent benign neural tumor types: neurofibroma, perineurioma, and schwannoma. METHODS: A model was developed, trained, and evaluated for classification using the ResNet-50 architecture, with a database of 30 whole-slide images stained in hematoxylin and eosin (106, 782 patches were generated from and divided among the training, validation, and testing subsets, with strategies to avoid data leakage). RESULTS: The model achieved an accuracy of 70% (64% normalized), and showed satisfactory results for differentiating two of the three classes, reaching approximately 97% and 77% as true positives for neurofibroma and schwannoma classes, respectively, and only 7% for perineurioma class. The AUROC curves for neurofibroma and schwannoma classes was 0.83%, and 0.74% for perineurioma. However, the specificity rate for the perineurioma class was greater (83%) than in the other two classes (neurofibroma with 61%, and schwannoma with 60%). CONCLUSION: This investigation demonstrated significant potential for proficient performance with a limitation regarding the perineurioma class (the limited feature variability observed contributed to a lower performance).

Artificial intelligence and radiomics in the diagnosis of intraosseous lesions of the gnathic bones: A systematic review.

BACKGROUND: The purpose of this systematic review (SR) is to gather evidence on the use of machine learning (ML) models in the diagnosis of intraosseous lesions in gnathic bones and to analyze the reliability, impact, and usefulness of such models. This SR was performed in accordance with the PRISMA 2022 guidelines and was registered in the PROSPERO database (CRD42022379298). METHODS: The acronym PICOS was used to structure the inquiry-focused review question "Is Artificial Intelligence reliable for the diagnosis of intraosseous lesions in gnathic bones?" The literature search was conducted in various electronic databases, including PubMed, Embase, Scopus, Cochrane Library, Web of Science, Lilacs, IEEE Xplore, and Gray Literature (Google Scholar and ProQuest). Risk of bias assessment was performed using PROBAST, and the results were synthesized by considering the task and sampling strategy of the dataset. RESULTS: Twenty-six studies were included (21 146 radiographic images). Ameloblastomas, odontogenic keratocysts, dentigerous cysts, and periapical cysts were the most frequently investigated lesions. According to TRIPOD, most studies were classified as type 2 (randomly divided). The F1 score was presented in only 13 studies, which provided the metrics for 20 trials, with a mean of 0.71 (±0.25). CONCLUSION: There is no conclusive evidence to support the usefulness of ML-based models in the detection, segmentation, and classification of intraosseous lesions in gnathic bones for routine clinical application. The lack of detail about data sampling, the lack of a comprehensive set of metrics for training and validation, and the absence of external testing limit experiments and hinder proper evaluation of model performance.

Deep learning applied to the histopathological diagnosis of ameloblastomas and ameloblastic carcinomas.

BACKGROUND: Odontogenic tumors (OT) are composed of heterogeneous lesions, which can be benign or malignant, with different behavior and histology. Within this classification, ameloblastoma and ameloblastic carcinoma (AC) represent a diagnostic challenge in daily histopathological practice due to their similar characteristics and the limitations that incisional biopsies represent. From these premises, we wanted to test the usefulness of models based on artificial intelligence (AI) in the field of oral and maxillofacial pathology for differential diagnosis. The main advantages of integrating Machine Learning (ML) with microscopic and radiographic imaging is the ability to significantly reduce intra-and inter observer variability and improve diagnostic objectivity and reproducibility. METHODS: Thirty Digitized slides were collected from different diagnostic centers of oral pathology in Brazil. After performing manual annotation in the region of interest, the images were segmented and fragmented into small patches. In the supervised learning methodology for image classification, three models (ResNet50, DenseNet, and VGG16) were focus of investigation to provide the probability of an image being classified as class0 (i.e., ameloblastoma) or class1 (i.e., Ameloblastic carcinoma). RESULTS: The training and validation metrics did not show convergence, characterizing overfitting. However, the test results were satisfactory, with an average for ResNet50 of 0.75, 0.71, 0.84, 0.65, and 0.77 for accuracy, precision, sensitivity, specificity, and F1-score, respectively. CONCLUSIONS: The models demonstrated a strong potential of learning, but lack of generalization ability. The models learn fast, reaching a training accuracy of 98%. The evaluation process showed instability in validation; however, acceptable performance in the testing process, which may be due to the small data set. This first investigation opens an opportunity for expanding collaboration to incorporate more complementary data; as well as, developing and evaluating new alternative models.

The use of deep learning state-of-the-art architectures for oral epithelial dysplasia grading: A comparative appraisal.

BACKGROUND: Dysplasia grading systems for oral epithelial dysplasia are a source of disagreement among pathologists. Therefore, machine learning approaches are being developed to mitigate this issue. METHODS: This cross-sectional study included a cohort of 82 patients with oral potentially malignant disorders and correspondent 98 hematoxylin and eosin-stained whole slide images with biopsied-proven dysplasia. All whole-slide images were manually annotated based on the binary system for oral epithelial dysplasia. The annotated regions of interest were segmented and fragmented into small patches and non-randomly sampled into training/validation and test subsets. The training/validation data were color augmented, resulting in a total of 81,786 patches for training. The held-out independent test set enrolled a total of 4,486 patches. Seven state-of-the-art convolutional neural networks were trained, validated, and tested with the same dataset. RESULTS: The models presented a high learning rate, yet very low generalization potential. At the model development, VGG16 performed the best, but with massive overfitting. In the test set, VGG16 presented the best accuracy, sensitivity, specificity, and area under the curve (62%, 62%, 66%, and 65%, respectively), associated with the higher loss among all Convolutional Neural Networks (CNNs) tested. EfficientB0 has comparable metrics and the lowest loss among all convolutional neural networks, being a great candidate for further studies. CONCLUSION: The models were not able to generalize enough to be applied in real-life datasets due to an overlapping of features between the two classes (i.e., high risk and low risk of malignization).

Machine learning for the prediction of toxicities from head and neck cancer treatment: A systematic review with meta-analysis.

INTRODUCTION: The aim of the present systematic review (SR) is to summarize Machine Learning (ML) models currently used to predict head and neck cancer (HNC) treatment-related toxicities, and to understand the impact of image biomarkers (IBMs) in prediction models (PMs). The present SR was conducted following the guidelines of the PRISMA 2022 and registered in PROSPERO database (CRD42020219304). METHODS: The acronym PICOS was used to develop the focused review question (Can PMs accurately predict HNC treatment toxicities?) and the eligibility criteria. The inclusion criteria enrolled Prediction Model Studies (PMSs) with patient cohorts that were treated for HNC and developed toxicities. Electronic database search encompassed PubMed, EMBASE, Scopus, Cochrane Library, Web of Science, LILACS, and Gray Literature (Google Scholar and ProQuest). Risk of Bias (RoB) was assessed through PROBAST and the results were synthesized based on the data format (with and without IBMs) to allow comparison. RESULTS: A total of 28 studies and 4,713 patients were included. Xerostomia was the most frequently investigated toxicity (17; 60.71 %). Sixteen (57.14 %) studies reported using radiomics features in combination with clinical or dosimetrics/dosiomics for modelling. High RoB was identified in 23 studies. Meta-analysis (MA) showed an area under the receiver operating characteristics curve (AUROC) of 0.82 for models with IBMs and 0.81 for models without IBMs (p value < 0.001), demonstrating no difference among IBM- and non-IBM-based models. DISCUSSION: The development of a PM based on sample-specific features represents patient selection bias and may affect a model's performance. Heterogeneity of the studies as well as non-standardized metrics prevent proper comparison of studies, and the absence of an independent/external test does not allow the evaluation of the model's generalization ability. CONCLUSION: IBM-featured PMs are not superior to PMs based on non-IBM predictors. The evidence was appraised as of low certainty.

Machine learning concepts applied to oral pathology and oral medicine: A convolutional neural networks' approach.

INTRODUCTION: Artificial intelligence models and networks can learn and process dense information in a short time, leading to an efficient, objective, and accurate clinical and histopathological analysis, which can be useful to improve treatment modalities and prognostic outcomes. This paper targets oral pathologists, oral medicinists, and head and neck surgeons to provide them with a theoretical and conceptual foundation of artificial intelligence-based diagnostic approaches, with a special focus on convolutional neural networks, the state-of-the-art in artificial intelligence and deep learning. METHODS: The authors conducted a literature review, and the convolutional neural network's conceptual foundations and functionality were illustrated based on a unique interdisciplinary point of view. CONCLUSION: The development of artificial intelligence-based models and computer vision methods for pattern recognition in clinical and histopathological image analysis of head and neck cancer has the potential to aid diagnosis and prognostic prediction.

The value of artificial intelligence for detection and grading of prostate cancer in human prostatectomy specimens: a validation study.

BACKGROUND: The Gleason grading system is an important clinical practice for diagnosing prostate cancer in pathology images. However, this analysis results in significant variability among pathologists, hence creating possible negative clinical impacts. Artificial intelligence methods can be an important support for the pathologist, improving Gleason grade classifications. Consequently, our purpose is to construct and evaluate the potential of a Convolutional Neural Network (CNN) to classify Gleason patterns. METHODS: The methodology included 6982 image patches with cancer, extracted from radical prostatectomy specimens previously analyzed by an expert uropathologist. A CNN was constructed to accurately classify the corresponding Gleason. The evaluation was carried out by computing the corresponding 3 classes confusion matrix; thus, calculating the percentage of precision, sensitivity, and specificity, as well as the overall accuracy. Additionally, k-fold three-way cross-validation was performed to enhance evaluation, allowing better interpretation and avoiding possible bias. RESULTS: The overall accuracy reached 98% for the training and validation stage, and 94% for the test phase. Considering the test samples, the true positive ratio between pathologist and computer method was 85%, 93%, and 96% for specific Gleason patterns. Finally, precision, sensitivity, and specificity reached values up to 97%. CONCLUSION: The CNN model presented and evaluated has shown high accuracy for specifically pattern neighbors and critical Gleason patterns. The outcomes are in line and complement others in the literature. The promising results surpassed current inter-pathologist congruence in classical reports, evidencing the potential of this novel technology in daily clinical aspects.

Robust pulmonary segmentation for chest radiography, combining enhancement, adaptive morphology and innovative active contours

Abstract Introduction Statistical data reveal that approximately 140 million radiological exams are performed annually in Brazil. These exams are designed to detect and to analyze fractures, caused by different types of trauma; as well as, to diagnose pathologies such as pulmonary diseases. For better visualization of those lesions or abnormalities, methods of image segmentation can be implemented. Such methods lead to the separation of the region of interest, which allows extracting the characteristics and anomalies of the desired tissue. However, the methods developed by researchers in this area still have restrictions. Consequently, we present an automatic pulmonary segmentation approach that overcomes these constraints. Methods This method is composed of a combination of Discrete Wavelet Packet Frame (DWPF), morphological operations and Gradient Vector Flow (GVF). The methodology is divided into four steps: Pre-processing - the original image is enhanced by discrete wavelet; Processing - where occurs a combination of the Otsu threshold with a series of morphological operations in order to identify the pulmonary object; Post-processing - an innovative form of using GVF improves the binary information of pulmonary tissue, and; Evaluation - the segmented images were evaluated for accuracy of detection the pulmonary region and border. Results The evaluation was carried out by segmenting 247 digital X-ray challenging images of the thorax human. The results show high for values of Overlap (97,63% ± 3.34%), and Average Contour Distance (0.69mm ± 0.95mm). Conclusion The results allow verifying that the proposed technique is robust and more accurate than other methods of lung segmentation, besides being a fully automatic method of lung segmentation.

Realistic deformable 3D numeric phantom for transcutaneous ultrasound

Abstract Introduction Numerical phantoms are important tools to design, calibrate and evaluate several methods in various image-processing applications, such as echocardiography and mammography. We present a framework for creating ultrasound numerical deformable phantoms based on Finite Element Method (FEM), Linear Isomorphism and Field II. The proposed method considers that the scatterers map is a property of the tissue; therefore, the scatterers should move according to the tissue strain. Methods First, a volume representing the target tissue is loaded. Second, parameter values, such as Young’s Modulus, scatterers density, attenuation and scattering amplitudes are inserted for each different regions of the phantom. Then, other parameters related to the ultrasound equipment, such as ultrasound frequency and number of transducer elements, are also defined in order to perform the ultrasound acquisition using Field II. Third, the size and position of the transducer and the pressures that are applied against the tissue are defined. Subsequently, FEM is executed and deformation is computed. Next, 3D linear isomorphism is performed to displace the scatterers according to the deformation. Finally, Field II is carried out to generate the non-deformed and deformed ultrasound data. Results The framework is evaluated by comparing strain values obtained the numerical simulation and from the physical phantom from CIRS. The mean difference between both phantoms is lesser than 10%. Conclusion The acoustic and deformation outcomes are similar to those obtained using a physical phantom. This framework led to a tool, which is available online and free of charges for educational and research purposes.

Atherosclerotic plaque characterization using plaque area variation in IVUS images during compression: a computational investigation

INTRODUCTION: The rupture of atherosclerotic plaques causes millions of death yearly. It is known that the kind of predominant tissue is associated with its dangerousness. In addition, the mechanical properties of plaques have been proved to be a good parameter to characterize the type of tissue, important information for therapeutic decisions. METHODS: Therefore, we present an alternative and simple way to discriminate tissues. The procedure relies on computing an index, the ratio of the plaque area variation of a suspecting plaque, using images acquired with vessel and plaques, pre and post-deformation, under different intraluminal pressure. Numerical phantoms of coronary cross-sections with different morphological aspects, and simulated with a range of properties, were used for evaluation. RESULTS: The outcomes provided by this index and a widely used one were compared, so as to measure their correspondence. As a result, correlations up to 99%, a strong agreement with Bland-Altman and very similar histograms between the two indices, have shown a good level of equivalence between the methods. CONCLUSION: The results demonstrated that the proposed index discriminates highly lipidic from fibro-lipidic and calcified tissues in many situations, as good as the widely used index, yet the proposed method is much simpler to be computed.

Segmentação do lúmen em imagens de IOCT usando Fuzzy Connectedness e Reconstrução Binária Morfológica / Lumen Segmentation in IOCT images using Fuzzy and Binary Morphological Reconstruction

No ano 2010, doenças cardiovasculares (CVD) causaram 33% do total das mortes no Brasil. Tomografia Ótica Coerente Intravascular (IOCT) é uma tecnologia que oferece imagens in vivo para detecção e monitoramento da progressão de CVD. O exame de IOCT permite mais precisão no diagnóstico; contudo, ainda é pequena a variedade de métodos quantitativos aplicados a IOCT na literatura, em comparação à outras modalidades relacionadas. Portanto neste trabalho é proposto um método de segmentação do lúmen, baseado em uma combinação de Fuzzy Connectedness, com múltiplas funções de afinidade, e Operações Morfológicas. As funções de afinidade usadas neste trabalho são: (I) Clássica, (II) Pesos Dinâmicos e (III) Bhattacharyya. Esta última é baseada no coeficiente de Bhattacharyya, utilizado habitualmente para speckle tracking. Primeiro, características não desejadas da imagem são atenuadas. Depois, informações da parede do vaso são obtidas utilizando Fuzzy Connectedness e um processo de binarização dinâmico. Finalmente, operações morfológicas são realizadas para melhorar o lúmen segmentado. Para avaliar o método proposto, um conjunto de 130 imagens advindas de humanos, porcos, e coelhos foram segmentadas e comparadas com seus respectivos "Gold Standards" feitos por especialistas. Uma média de verdadeiros positivos (TP%) = 98,08 e de falsos positivos (FP%) = 2,34 foram obtidas. Com isso, o método proposto resultou em uma maior eficácia do que os estudos publicados anteriormente, encorajando seu uso.

In 2010 cardiovascular disease (CVD) caused 33% of the total deaths in Brazil. Intravascular Optical Coherent Tomography (IOCT) is an imaging technology that provides in vivo detection and monitoring of the progression of coronary heart disease. IOCT exam allows more accurate diagnoses; nonetheless, the set of quantitative methods applied to IOCT in the literature is small compared to other related modalities. Therefore, the proposed approach presents a lumen segmentation method, based on a combination of Fuzzy Connectedness, with multiple affinity functions, and Morphological Operations. The affinity functions used in this work are: (I) classical, (II) Dynamic weights (III) Bhattacharyya. The latter is based on the Bhattacharyya coefficient, commonly used for speckle tracking. Firstly, unwanted features of the image are attenuated. Then, vessel-wall information is obtained using Fuzzy Connectedness and dynamic binarization process. Finally, morphological operations are performed to improve the segmented lumen. To evaluate the proposed method, a set of 130 images from humans, pigs and rabbits were segmented and compared to their corresponding gold standard made by experts. An average of true positive (TP%) = 98.08, and false positive (FP%) = 2.34 were obtained. Hence, the use of the proposed method is suggested since it has yielded higher efficiency than previously published studies.

An approach to automatically segment the media-adventitia borders in IVUS / Uma associação de técnicas para segmentação automática da fronteira da média-adventícia em IVUS

Por ser capaz de mostrar aspectos morfológicos e patológicos de ateroscleroses, o Ultrassom Intravascular (IVUS) se tornou uma das modalidades de imagens médicas mais confiáveis e empregadas em intervenções cardíacas. As características de sua imagem aumentam as chances de um bom diagnóstico, resultando em terapias mais precisas. O estudo de segmentação da fronteira média-adventícia, dentre muitas aplicações, é importante para o aprendizado das propriedades mecânicas e determinação de algumas medidas específicas (raio, diâmetro, etc.) em vasos e placas. Neste trabalho, uma associação de técnicas de processamento de imagens está sendo proposta para atingir alta acurácia na segmentação da borda média-adventícia. Para tanto, foi feita uma combinação das seguintes técnicas: Redução do Speckle por Difusão Anisotrópica (SRAD), Wavelet, Otsu e Morfologia Matemática. Primeiramente, é usado SRAD para atenuar os ruídos speckle. Posteriormente, é executada Transformada Wavelet para extração das características dos vasos e placas. Uma versão binarizada dessas características é criada na qual o limiar ótimo é definido por Otsu. Finalmente, é usada Morfologia Matemática para obtenção do formato da adventícia. O método proposto é avaliado ao segmentar 100 imagens de alta complexidade, obtendo uma média de Verdadeiro Positivo (TP(%)) = 92,83 ± 4,91, Falso Positivo (FP(%)) = 3,43 ± 3,47, Falso Negativo (FN(%)) = 7,17 ± 4,91, Máximo Falso Positivo (MaxFP(mm)) = 0,27 ± 0,22, Máximo Falso Negativo (MaxFN(mm)) = 0,31 ± 0,2. A eficácia do nosso método é demonstrada, comparando este resultado com outro trabalho recente na literatura.

By being able to show morphological and pathological aspects of atherosclerosis, the Intravascular Ultrasound (IVUS) be¬came one of the most reliable and employed medical imaging modality in cardiac interventions. Its image characteristics in¬crease the chances of a good diagnostic, resulting in a precise therapy. The study of media-adventitia borders segmentation in IVUS, among many applications, is important for learning about the mechanical properties and determining some specific measurements (radius, diameter, etc.) in vases and plaques. An approach is proposed to achieve high accuracy in media-adventitia borders segmentation, by making a combination of different image processing operations: Speckle Reducing Anisotropic Diffusion (SRAD), Wavelet, Otsu and Mathematical Morphology. Firstly, SRAD is applied to attenuate the speckle noise. Next, the vessel and plaque features are extracted by performing Wavelet Transform. Optimal thresholding is car¬ried out by Otsu method to create a binarized version of these features. Then, Mathematical Morphology operations are used to obtain an adventitia shape. The proposed approach is evaluated by segmenting 100 challenging images, obtaining an average of True Positive (TP(%)) = 92.83 ± 4.91, False Positive (FP(%)) = 3.43 ± 3.47, False Negative (FN(%)) = 7.17 ± 4.91, Max False Positive (MaxFP(mm)) = 0.27 ± 0.22, Max False Negative (MaxFN(mm)) = 0.31 ± 0.2. The effectiveness of our approach is demonstrated by comparing this result with another recent work in the literature.