Identification of Colon Cancer Using Multi-Scale Feature Fusion Convolutional Neural Network Based on Shearlet Transform

Abstract

Colon cancer identification is of great significance in medical diagnosis. Real-time, objective and accurate inspection results will facilitate medical professionals to explore symptomatic treatment promptly. However, the existing methods depend on hand-crafted features which require extensive professional expertise and long inspection period. Therefore, we propose a multi-scale feature fusion convolutional neural network (MFF-CNN) based on shearlet transform to identify histopathological image of colon cancer. The characteristic of the framework is the shearlet coefficients of histopathological image in multiple decomposition scales were extracted as supplementary feature which were also fed to the network with the original pathological image. After feature learning and feature fusion, the MFF-CNN based on shearlet transform can achieve the identification accuracy of 96% and average F-1 score of 0.9594 for colorectal adenocarcinoma epithelium (TUM) and normal colon mucosa (NORM). The false negative rate and false positive rate can be reduced to 5.5% and 2.5%, respectively. The superior performance of the network opens a new perspectives for real-time, objective and accurate diagnosis of cancer.

AI-based Carcinoma Detection and Classification Using Histopathological Images: A Systematic Review

“…• Pretrained networks and deep networks: It is evident in this article that out of 101 research papers, vast majority of contributions by deep networks were Gaussian blur smoothing [74,76] k-means [67,72], Graph-based [70], Dense UNet [70], Region growing [71,72], Thresholding [74,76,81], CNN [79] Zoom-out features [67], ConvNet features [65,67,76,99], Object graph features [72], Topological features [65], Various texture features [73,79,80], Morphology features [75,79,80] Transfer learning models [73,80,83], Machine learning algorithms [65,67,72,76,80,99], Custom CNN [69,74,78,80], Quasi-supervised learning [73], Neural network [66,74,81], Ensemble CNN [77] Prostate…”

“…Instead of single-center dataset, the model provided better accuracy of 99.13% when datasets were combined. Liang et al [78] presented a method to identify histopathological images of ADC using multi-scale feature fusion CNN-based shearlet transform technique. They achieved detection accuracy of 96% for 8,000 image patches in the training dataset.…”

Machine Learning-based Lung and Colon Cancer Detection using Deep Feature Extraction and Ensemble Learning

“…It demonstrated that hybrid deep learning models may be used to evaluate a reliable cancer diagnosis approach. On the other hand, (Liang et al, 2020) developed a shearlet transform-based multi-scale feature fusion (MFF-CNN) model to recognize histopathology images of colon cancer. It achieved a 96 percent detection performance for colorectal after feature learning and feature fusion.…”

“…A number of papers have been conducted in the literature, each of which is predicated on a distinct strategy for detecting cancer (Hatuwal and Thapa, 2020) (Tasnim et al, 2021) (Liang et al, 2020) (Chen et al, 2021).…”

Fully Automatic Model Based on SE-ResNet for Bone Age Assessment

“…Recently, the success of CNNs for medical image processing has been reported [18][19][20]. Deep learning methods based on CNNs have been suggested for image preprocessing and feature extraction in BAA tasks, such as fine-tuned CNNs [10], methods based on visual geometry groups (VGGs) [11], UNets for segmentation [12,13], deep residual network (ResNet)-based models [14], and CNNs with attention mechanisms [15].…”