Colon cancer heavily changes the composition of human genes (expressions). The deviation in the chemical composition of genes can be exploited to automatically diagnose colon cancer. The major challenge in the analysis of human gene based datasets is their large dimensionality. Therefore, efficient techniques are needed to select discerning genes. In this research article, we propose a novel classification technique that exploits the variations in gene expressions for classifying colon gene samples into normal and malignant classes, and quite intelligently tackles the larger dimensionality of gene based datasets. Previously individual feature selection techniques have been used for selection of discerning gene expressions, however, their performance is limited. In this research study, we propose a feed forward gene selection technique, wherein, two feature selection techniques are used one after the other. The genes selected by the first technique are fed as input to the second feature selection technique that selects genes from the given gene subset. The selected genes are then classified by using linear kernel of support vector machines (SVM). The feed forward approach of gene selection has shown improved performance. The proposed technique has been tested on three standard colon cancer datasets, and improved performance has been observed. It is observed that feed forward method of gene selection substantially reduces the size of gene based datasets, thereby reducing the computational time to a great extent. Performance of the proposed technique has also been compared with existing techniques of colon cancer diagnosis, and improved performance has been observed. Therefore, we hope that the proposed technique can be effectively used for diagnosis of colon cancer.

This work investigates the application of Computer Vision to the problem of the automated counting and measuring of crabs and lobsters onboard fishing boats. The aim is to provide catch count and measurement data for these key commercial crustacean species. This can provide vital input data for stock assessment models, to enable the sustainable management of these species. The hardware system is required to be low-cost, have low-power usage, be waterproof, available (given current chip shortages), and able to avoid over-heating. The selected hardware is based on a Raspberry Pi 3A+ contained in a custom waterproof housing. This hardware places challenging limitations on the options for processing the incoming video, with many popular deep learning frameworks (even light-weight versions) unable to load or run given the limited computational resources. The problem can be broken into several steps: (1) Identifying the portions of the video that contain each individual animal; (2) Selecting a set of representative frames for each animal, e.g, lobsters must be viewed from the top and underside; (3) Detecting the animal within the frame so that the image can be cropped to the region of interest; (4) Detecting keypoints on each animal; and (5) Inferring measurements from the keypoint data. In this work, we develop a pipeline that addresses these steps, including a key novel solution to frame selection in video streams that uses classification, temporal segmentation, smoothing techniques and frame quality estimation. The developed pipeline is able to operate on the target low-power hardware and the experiments show that, given sufficient training data, reasonable performance is achieved.

Crabs and lobsters are valuable crustaceans that contribute enormously to the seafood needs of the growing human population. This paper presents a comprehensive analysis of single- and multi-stage object detectors for the detection of crabs and lobsters using images captured onboard fishing boats. We investigate the speed and accuracy of multiple object detection techniques using a novel dataset, multiple backbone networks, various input sizes, and fine-tuned parameters. We extend our work to train lightweight models to accommodate the fishing boats equipped with low-power hardware systems. Firstly, we train Faster R-CNN, SSD, and YOLO with different backbones and tuning parameters. The models trained with higher input sizes resulted in lower frames per second (FPS) and vice versa. The base models were highly accurate but were compromised in computational and run-time costs. The lightweight models were adaptable to low-power hardware compared to the base models. Secondly, we improved the performance of YOLO (v3, v4, and tiny versions) using custom anchors generated by the k-means clustering approach using our novel dataset. The YOLO (v4 and it’s tiny version) achieved mean average precision (mAP) of 99.2% and 95.2%, respectively. The YOLOv4-tiny trained on the custom anchor-based dataset is capable of precisely detecting crabs and lobsters onboard fishing boats at 64 frames per second (FPS) on an NVidia GeForce RTX 3070 GPU. The Results obtained identified the strengths and weaknesses of each method towards a trade-off between speed and accuracy for detecting objects in input images.

Vascular changes in small vessels can be observed through Flair MR images called white matter hyperintensities (WMHs). WMHs are associated with cerebral small vessel disease, aging, dementia, and stroke. Quantification of WHMs is important for diagnosis, prognosis, monitoring of patients, and research studies. Manual segmentation of WHMs is a time-consuming task and is subjective to the observer. That's why we need an automated segmentation method. WHMs' segmentation is a challenging task due to their heterogeneous characteristics and coexistence with other similar-appearing structures. We proposed an architecture that uses residual blocks in a U-Net-based network to segment the WHMs from T1-weighted, Flair images of brains. In this study, we used the MICCIA White Matter Hyperintensities Challenge 2017 Dataset to train and test the proposed model. We evaluated the proposed method using the standard MWM challenge in 2017 evaluation measures and achieved better results than the state-of-the-art technique. [1]