Vertebra Segmentation for Clinic CT Image using Mask R-CNN

EasyChair Preprint 2520

Abstract

Spine disease is a growing problem in modern society and has been debilitating for every age-group. Research has shown that more than 266 million people are facing degenerative spine disease and low back pain. CT scanning is a fast, painless, non-invasive diagnostic imaging modality that provides high spatial accuracy in obtaining the 3D structure of the vertebral. However, the clinic CT image might not cover the whole spine and not has the same field of view in real-life situations. Henceforth, this project aims to create and validate an automatic method that can detect, locate, and classify each vertebra from the partial field of view using deep learning. Mask R-CNN is a deep neural network aimed to solve the instance segmentation problem in machine learning or computer vision, and the bounding boxes, classes, and masks are used to identify each vertebra. This auto-detection method has been successfully implemented on open source dataset which has been used on Computational Spine Imaging (CSI 2014). The dataset was physically chosen by a radiologist with an eight-year-long time of involvement based on thoracic and lumbar spine column scans, and the data of twenty patients were collected using CT protocol. The accuracy of the verte-bra mask on 210 test images has been increased up to 99.9% DICE Coefficient in Mask R-CNN compare with 69.2% DICE Coefficient in U-Net.

Keyphrases: Mask R-CNN, Partial Clinic CT Image, Vertebra Segmentation, spine disease

@booklet{EasyChair:2520,
  author    = {Renjie Wang and Jennifer Hui Yi Voon and Da Ma and Setareh Dabiri and Karteek Popuri and Mirza Faisal Beg},
  title     = {Vertebra Segmentation for Clinic CT Image using Mask R-CNN},
  howpublished = {EasyChair Preprint 2520},
  year      = {EasyChair, 2020}}