Explainable AI-driven model for gastrointestinal cancer classification

Frontiers in Medicine, Journal Year: 2024, Volume and Issue: 11

Published: April 15, 2024

Although the detection procedure has been shown to be highly effective, there are several obstacles overcome in usage of AI-assisted cancer cell clinical settings. These issues stem mostly from failure identify underlying processes. Because diagnosis does not offer a clear decision-making process, doctors dubious about it. In this instance, advent Explainable Artificial Intelligence (XAI), which offers explanations for prediction models, solves AI black box issue. The SHapley Additive exPlanations (SHAP) approach, results interpretation model predictions, is main emphasis work. intermediate layer study was hybrid made up three Convolutional Neural Networks (CNNs) (InceptionV3, InceptionResNetV2, and VGG16) that combined their predictions. KvasirV2 dataset, comprises pathological symptoms associated cancer, used train model. Our yielded an accuracy 93.17% F1 score 97%. After training model, we use SHAP analyze images these groups provide explanation decision affects prediction.

Language: Английский

---

Enhancing Ulcerative Colitis Diagnosis: A Multi-Level Classification Approach with Deep Learning

Computer Modeling in Engineering & Sciences, Journal Year: 2024, Volume and Issue: 140(1), P. 1129 - 1142

Published: Jan. 1, 2024

The evaluation of disease severity through endoscopy is pivotal in managing patients with ulcerative colitis, a condition significant clinical implications.However, endoscopic assessment susceptible to inherent variations, both within and between observers, compromising the reliability individual evaluations.This study addresses this challenge by harnessing deep learning develop robust model capable discerning discrete levels severity.To initiate endeavor, multi-faceted approach embarked upon.The dataset meticulously preprocessed, enhancing quality discriminative features images contrast limited adaptive histogram equalization (CLAHE).A diverse array data augmentation techniques, encompassing various geometric transformations, leveraged fortify dataset's diversity facilitate effective feature extraction.A fundamental aspect involves strategic incorporation transfer principles, modified ResNet-50 architecture.This augmentation, informed domain expertise, contributed significantly model's classification performance.The outcome research endeavor yielded highly promising model, demonstrating an accuracy rate 86.85%, coupled recall 82.11% precision 89.23%.

Language: Английский

---

A Framework for Multi-Grade Classification of Ulcerative-Colitis Using Deep Neural Networks

Published: Nov. 17, 2023

Endoscopic disease severity assessment is a critical component in the management of ulcerative colitis patients. evaluation, on other hand, suffers from significant intra-observer and inter-observer differences, reducing reliability individual assessments. As result, we set out to create deep-learning model capable distinguishing between distinct endoscopic levels. Initially, preprocessed dataset then applied data augmentations images using various geometric transformations. Subsequently, have utilized transfer learning concept by applying modified ResNet-50 stacking additional layers which further improves classification performance. Our proposed achieved an accuracy 84.21%, 81.06% recall, 88.33% precision.

Language: Английский

---

A hybrid features fusion-based framework for classification of breast micronodules using ultrasonography

BMC Medical Imaging, Journal Year: 2024, Volume and Issue: 24(1)

Published: Sept. 20, 2024

Breast cancer is one of the leading diseases worldwide. According to estimates by National Cancer Foundation, over 42,000 women are expected die from this disease in 2024.

Language: Английский

---

Enhancing Colonoscopy Image Quality with CLAHE in the GASTROLAB Dataset

Published: Dec. 21, 2023

This research focuses on improving image quality in the context of colonoscopy, a critical procedure for diagnosing gastrointestinal conditions such as colorectal polyps, which can lead to cancer. The study highlights challenges posed by large volume frames generated during colonoscopy procedures, necessitates use automated systems detect anomalies. Despite fact that some datasets contain low-quality images, it emphasizes importance various disease and machine learning. primary goal this work is improve GASTROLAB dataset, emphasizing significance accurate results derived from high-quality data. looks into enhancement techniques, with focus Contrast Limited Adaptive Histogram Equalization (CLAHE), produces superior metrics like Peak Signal Noise Ratio (PSNR) and), Structural Similarity Index (SSIM), are 32.26 0.912, respectively. findings demonstrate utility PSNR SSIM assessment tools, while also clinical validation expert judgement medical evaluation. concludes support CLAHE preferred method quality, particularly when visualizing small polyps colonoscopy.

Language: Английский

---