Figure - available from: Evidence-based Complementary and Alternative Medicine
This content is subject to copyright. Terms and conditions apply.
Source publication
Early and automatic detection of colorectal tumors is essential for cancer analysis, and the same is implemented using computer-aided diagnosis (CAD). A computerized tomography (CT) image of the colon is being used to identify colorectal carcinoma. Digital imaging and communication in medicine (DICOM) is a standard medical imaging format to process...
Citations
... The authors of [27] gained a 95.79% F1-score and 91.62% Jaccard Index (Table 1), while the authors of [30] received the F1-score of 91.57%, an accuracy of 98.82%, a sensitivity of 98.28%, and a specificity of 98.68% (Table 2), from a dataset [31] known as The Cancer Imaging Archive in the context of CT Colonography [32]. ...
... ReLU activation functions accelerate the training process by mitigating the vanishing gradient problem and enabling faster convergence compared to saturating nonlinearities. The deep and complex architecture of AlexNet may reduce model interpretability, making it challenging to understand how the network makes predictions or which features are essential for classification [30]. Understanding the internal representations learned by the network can be more challenging in architectures with many layers and parameters. ...
... Understanding the internal representations learned by the network can be more challenging in architectures with many layers and parameters. Using the following dataset [34], the authors of [30] got a 44% precision and 63.78% sensitivity, while for the work [33], this model obtained the F1-score value bigger than 80%, precision of 88.3%, and a sensitivity of 74.8%. This architecture is excellent in classifying images using hierarchical class information [35]. ...
The third most prevalent cancer nowadays is colorectal cancer. Colonoscopy is an important procedure in the stage of detection of polyps' malignancy because it helps in early identification and establishes effective therapy.
This paper explores specific deep-learning architectures for the binary classification of colorectal polyps and considers the evaluation of their premalignancy risk.
The main scope is to create a custom-based deep learning architecture that classifies adenomatous, hyperplastic, and serrated polyps' samples into benign and premalignant
based on images from the colonoscopic dataset. Each image's output is modified through masked autoencoders which enhance the classification performance of the proposed model, called Bionnica.
From the four evaluated state-of-the-art deep learning models (ZF NET, VGG-16, AlexNet, and ResNet-50), our experiments showed that ResNet-50 and ZF NET are most accurate (above 84%), with ResNet-50 excelling at indicating patients with premalignant colorectal polyps (above 92%). ZF NET is the fastest at handling 700 images. Our proposed deep learning model, Bionnica, is more performant than ZF NET and provides an efficient classification of colorectal polyps given its simple structure. The advantage of our model comes from the custom enhancement interpretability with a rule-based layer that guides the learning process and supports medical personnel in their decisions.
... Therefore, automatic segmentation methods have been considered recently. Automatic segmentation models could be more accurate, less time-consuming, and applicable to large datasets (103,104). Using radiomics tumor imaging might cut expenses and individual errors compared to manual approaches. Using ResNet50 and ResNet152-V2 deep learning models may increase the detection accuracy of malignant ulcers (105). ...
With a high rate of morbidity and mortality, colorectal cancer (CRC) ranks third in mortality among cancers. By analyzing the texture properties of images and quantifying the heterogeneity of tumors, radiomics and radiogenomics are non-invasive methods to determine the biological properties of CRC. Recently, several articles have discussed the application of radiomics in different aspects of CRC. Therefore, given the large amount of data published, this review aims to discuss how radiomics can be used for distinguishing benign and malignant colorectal lesions, diagnosing, staging, predicting prognosis and treatment response, and predicting lymph node and hepatic metastasis of CRC, based on radiomic features extracted from magnetic resonance imaging (MRI), computed tomography (CT), esophageal ultrasonography (EUS), and positron emission tomography-CT (PET-CT). Challenges in bringing radiomics to clinical application and future solutions have also been discussed. With the progress made in radiomics and the application of deep and machine learning in this area, radiomics can become one of the main tools for the personalized management of CRC patients shortly.
... Therefore, automatic segmentation methods have been considered recently. Automatic segmentation models could be more accurate, less time-consuming, and applicable to large datasets (103,104). Using radiomics tumor imaging might cut expenses and individual errors compared to manual approaches. Using ResNet50 and ResNet152-V2 deep learning models may increase the detection accuracy of malignant ulcers (105). ...
With a high rate of morbidity and mortality, colorectal cancer (CRC) ranks third in mortality among cancers. By analyzing the texture properties of images and quantifying the heterogeneity of tumors, radiomics and radiogenomics are non-invasive methods to determine the biological properties of CRC. Recently, several articles have discussed the application of radiomics in different aspects of CRC. Therefore, given the large amount of data published, this review aims to discuss how radiomics can be used for distinguishing benign and malignant colorectal lesions, diagnosing, staging, predicting prognosis and treatment response, and predicting lymph node and hepatic metastasis of CRC, based on radiomic features extracted from magnetic resonance imaging (MRI), computed tomography (CT), esophageal ultrasonography (EUS), and positron emission tomography-CT (PET-CT). Challenges in bringing radiomics to clinical application and future solutions have also been discussed. With the progress made in radiomics and the application of deep and machine learning in this area, radiomics can become one of the main tools for the personalized management of CRC patients shortly.
... They compared the findings of these models to those produced by machine learning techniques. To segment the colon and polyps from CT images, a deep CNN-based residual network technique provided by Akilandeswari et al. [15] has been adopted over the 2D CT images. The residual stack block and the short skip nuance have been implemented in the hidden layers to preserve the spatial data. ...
Cancer is the second biggest cause of death worldwide, accounting for one of every six deaths. On the other hand, early detection of the disease significantly improves the chances of survival. The use of Artificial Intelligence (AI) to automate cancer detection might allow us to evaluate more cases in less time. In this research, AI-based deep learning models are proposed to classify the images of eight kinds of cancer, such as lung, brain, breast, and cervical cancer. This work evaluates the deep learning models, namely Convolutional Neural Networks (CNN), against classifying images with cancer traits. Pre-trained CNN variants such as MobileNet, VGGNet, and DenseNet are employed to transfer the knowledge they learned with the ImageNet dataset to detect different kinds of cancer cells. We use Bayesian Optimization to find the suitable values for the hyperparameters. However, transfer learning could make it so that models can no longer classify the datasets they were initially trained. So, we use Learning without Forgetting (LwF), which trains the network using only new task data while keeping the network’s original abilities. The results of the experiments show that the proposed models based on transfer learning are more accurate than the current state-of-the-art techniques. We also show that LwF can better classify both new datasets and datasets that have been trained before.
Despite therapeutic advancements, disease-free survival and overall survival of patients with locally advanced rectal cancer have not improved in most trials as a result of distant metastases. For treatment decision-making, both long-term oncologic outcomes and impact on quality-of-life indices should be considered (for example, bowel function). Total neoadjuvant therapy (TNT), comprised of chemotherapy and radiotherapy or chemoradiotherapy, is now a standard treatment approach in patients with features of high-risk disease to prevent local recurrence and distant metastases. In selected patients who have a clinical complete response, subsequent surgery might be avoided through non-operative management, but patients who do not respond to TNT have a poor prognosis. Refined molecular characterization might help to predict which patients would benefit from TNT and non-operative management. Specifically, integrated analysis of spatiotemporal multi-omics using artificial intelligence and machine learning is promising. Three prospective trials of TNT and non-operative management in Japan, the USA and Germany are collaborating to better understand drivers of response to TNT. Here, we address the future direction for TNT.
This study presents a robust framework for the classification of brain tumors, beginning with meticulous data curation from 233 patients. The dataset comprises a diverse range of T1-weighted contrast-enhanced images, encompassing meningioma, glioma, and pituitary tumor types. Rigorous organization, pre-processing, and augmentation techniques are applied to optimize model training. The proposed self-adaptive model incorporates a cutting-edge algorithm, leveraging Adaptive Contrast Limited Histogram Equalization (CLAHE) and Self-Adaptive Spatial Attention. CLAHE enhances grayscale images by tailoring contrast to the unique characteristics of each region. The Self-Adaptive Spatial Attention, implemented through an Attention Layer, dynamically assigns weights to spatial locations, thereby enhancing sensitivity to critical brain regions. The model architecture integrates transfer learning models, including DenseNet169, DenseNet201, ResNet152, and InceptionResNetV2, contributing to its robustness. DenseNet169 serves as a feature extractor, capturing hierarchical features through pre-trained weights. Model adaptability is further enriched by components such as batch normalization, dropout, layer normalization, and an adaptive learning rate strategy, mitigating overfitting and dynamically adjusting learning rates during training. Technical details, including the use of the Adam optimizer and softmax activation function, underscore the model's optimization and multi-class classification capabilities. The proposed model, which amalgamates transfer learning and adaptive mechanisms, emerges as a powerful tool for brain tumor detection and classification in medical imaging. Its nuanced comprehension of brain tumor images, facilitated by self-adaptive attention mechanisms, positions it as a promising advancement in computer-aided diagnosis in neuroimaging. Leveraging DenseNet201 with a self-adaptive mechanism, the model surpasses previous methods, achieving an accuracy of 94.85%, precision of 95.16%, and recall of 94.60%, showcasing its potential for enhanced accuracy and generalization in the challenging realm of medical image analysis.
In the past decade, significant progress has been made in the fields of artificial intelligence, machine learning, and deep learning (DL). These advancements have opened up wide applications and opportunities in the medical field. Colorectal cancer (CRC) has gained substantial interest from researchers due to its ranking as the third most prevalent cancer type after breast and lung cancer, affecting around 10% of all cancer patients globally each year. It is the second leading cause of cancer-related death worldwide, making the development of efficient computer-assisted methods for its detection, prediction, and treatment crucial. There are modalities used for colorectal cancer screening and detection such as colonoscopy images, biopsy samples, biomarker data, blood samples, CT scan, MRI, ultrasound, PET, and microbial data. The advancement of technology has made deep learning an attractive choice for fast and effective detection, segmentation, and prediction of diseases through image analysis. This technology has the potential to assist and empower medical professionals in making timely and informed decisions. Deep learning has proven to be highly effective when ample high-quality features and large datasets are available. However, one of the main challenges in using deep learning for medical image analysis is the limited availability of datasets from medical centers. This chapter provides an overview of DL-based models and their application in detecting and predicting CRC from various modalities. On the SCPolyps dataset the OEM model achieved training and test accuracy of 98% and 96% respectively.
