Deep Active Contour-Based Capsule Network for Medical Image Segmentation

ABSTRACT

Medical Image Segmentation (MIS) has important ramifications for the whole of medical diagnostics. Despite the unparalleled success that various Deep Learning (DL) techniques had in image segmentation, there are setbacks with them. Generally, these setbacks arise from two of the most profound indicators of performance in DL, namely network architecture and loss function, both of which are being dealt with in the proposed method to improve the dice score on the task of MIS. To deal with the architecture part of our work, we have used Capsules that have proved to preserving more image details and indicating the spatial relationships found locally and globally, facilitating the detection of global features from the local features extracted, and hence we propose a modified capsules network for the proposed segmentation task. The loss function used in the proposed work draws inspiration from the active contours model that tends to incorporate external forces, regional information and other functions of choices in the working of segmentation extraction and treats this task as a continuous curve evolution and energy minimization problem. In our experimentation, we have used a brain tumor segmentation dataset for the performance comparison using Dice-score and mean-IoU as performance metrics and the results obtained are very optimistic.

KEYWORDS:

• Active contour models
• Capsules network
• Convolutional neural networks
• Deep learning
• Medical image segmentation

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Narasimha Reddy Soora

Narasimha Reddy Soora received his post-graduate degrees (PhD(CSE)) from VNIT, Nagpur, Maharashtra, India, in 2017 and (MTech(CSE)) from JNTU, Hyderabad, Telangana, India, in 2007 and his undergraduate degree (BE(CSE)) from Osmania University, Hyderabad, Telangana, India in 1999. He has 20 years of experience, 9 years of which were in the IT industry and 11 years as an academic. His research interests are in image processing, machine learning, and deep learning. He is a life member of ISTE, CSI, India and a Fellow of IETE, India. Email: [email protected]

Ehsan Ur Rahman Mohammed

Mohammed Ehsan Ur Rahman received his BTech degree in computer science and engineering from Kakatiya Institute of Technology and Science, Warangal, Telangana State, India. His research interests are in computer vision, machine learning, and image processing with an emphasis on image retrieval, object detection and image classification alongside a developing interest in reinforcement learning. He is a member of ACM and ACM's SIGAI.

Sharfuddin Waseem Mohammed

Mohammed Sharfuddin Waseem is currently pursuing his PhD (CSE) from NIT Tiruchirappalli, Tamil Nadu, India, and is an assistant professor, Dept of CSE, KITSW, Warangal, Telangana State, India. He has 11 years of experience as an academic. He does research in image processing, deep learning with an emphasis on object detection. He is also active in doing projects in java and web development. Apart from research and teaching, he involves himself in creating innovative solutions and has received grants to build projects which can be of great service to society. Email: [email protected]

N. C. Santosh Kumar

N C Santosh Kumar received his MTech in software engineering from Kakatiya Institute of Technology and Science, Kakatiya University, Warangal. Currently, he is an assistant professor in the Department of CSE at Kakatiya Institute of Technology and Science and has more than 12 years of teaching experience. His research interest is biomedical image processing and he published 4 research articles in international journals and 1 research article in international conference, and 1 research article national journal to date in biomedical image processing, spatial data mining, and software testing. Currently, he is a research scholar at GIT, GITAM University, Vishakhapatnam, AP, India. Email: [email protected]