Deep Learning in Medical Image Classification and Object Detection: A Survey

Priyanka Gupta; Shikha Gupta

Abstract

Deep learning methods have demonstrated superior performance in the area of computer vision, speech recognition, natural language processing, healthcare, and many more. Convolutional neural networks (CNNs) are a class of deep learning methods that have the ability to learn from raw data available as images, audio, or text. CNNs have become a powerful tool for variety of pattern recognition tasks due to the availability of abundant data and GPU-based training. Usually, a CNN is designed with the following: (1) Convolution layers, (2) Pooling layers, and (3) Fully connected layers. Convolutional layers use convolution filters to extract the low-level features (like edges, circles) and high-level features (like objects, texture) from the input. Pooling layers are interleaved in between the convolution layers to reduce the input dimension for the subsequent layers. A fully connected layer makes use of extracted features from the pooling or convolutional layer and maps them to the final output, such as in the case of classification. In the domain of medical imaging analysis, deep learning methods are rapidly becoming state-of-the-art, achieving magnificent performances in many medical applications amid the challenges of unavailability of large amounts of medical data, and lack of annotated data. In the present work, we seek to review the application of deep learning approaches in the domain of medical imaging. We highlight the impact of deep learning methods with respect to two key areas: image classification and object detection, and give comprehensive summaries of findings in these areas. Future research directions and solutions are also explored.

Deep Learning in Medical Image Classification and Object Detection: A Survey

Abstract

Read more of it

Publish your article

Abstract

Related articles

AI-Based Deepfake Voice Detection Model for Hindi Language

DEVELOPING A DRIVER DROWSINESS DETECTION SYSTEM (DDDS)

A Study on CNN-Based Image Mosaicing

Advanced Classification of Diabetic Retinopathy Using GAN and Deep Residual CNN Model

Vani-Adapt: A Zero-Shot Accent Trans-Adaptation Framework for Robust Indic Speech Recognition

Enhancing Facial Expression Analysis Using CNNs: Insights from ML and Deep Learning Comparisons

Read more of it

Publish your article