Glioma resection cavity segmentation on neuroimaging using AI-based fuzzy enhancement with modified UNet (pp.338-354)
Chandana Kuntala¹, Sristi², Koyel Datta Gupta³, Amita Dev¹, Deepak Kumar Sharma^1,*, Deepak Gupta⁴
https://doi.org/10.30546/1683-6154.22.3.2023.338

²Department of Computer Science and Engineering, Indira Gandhi Delhi Technical University for Women, New Delhi, India, e-mail: sristi0108@gmail.com

³Department of Computer Science and Engineering, Maharaja Surajmal Institute of Technology, New Delhi, India, e-mail: koyel.dg@msit.in

⁴Department of Computer Science and Engineering, Maharaja Agrasen Institute of Technology (GGSIPU) Sector-22, Rohini, Delhi 110086, India, e-mail: deepakgupta@mait.ac.in

 ^*Corresponding author’s e-mail: : dk.sharma1982@yahoo.com

Abstract. This paper presents a three-tier fuzzy-infused segmentation pipeline for Low-Grade Glioma Resection Cavity (LGGRC). A Fuzzy-c-means based segregation technique and Fuzzy Inference System based enhancement method are proposed followed by a robust and lightweight architecture for post-operative ultrasound (US) image segmentation. This pipeline provides an end-to-end framework for medical image processing and the dichotomization of LGGRC from post-operative ultrasound images. Comprehensive experiments on our proposed method and its alternatives demonstrate that the suggested technique outperforms the state-of-the-art medical image segmentation approaches on the publicly available RESECT dataset. The proposed model yields better segmentation results by achieving Binary Accuracy (BAcc) of 0.9978, Jaccard Similarity Coefficient (mIoU) of 0.8462, Dice Sorenson Coefficient (DSC) of 0.9162, F1 Score (f1) of 0.9203. This indicates the efficiency of our method for generating,precise and consistent automated segmentations of US medical images.

Keywords: Artificial Intelligence, Healthcare, Fuzzy-c-Means, Fuzzy Inference System, UNet, Image Segmentation, Glioma Resection Cavity, Neuroimaging, Post-Operative Ultrasound.

AMS Subject Classification: 92B20, 92C55.