Revisiting segmentation of lung tumors from CT images

Farhanaz Farheen; Md Salman Shamil; Nabil Ibtehaz; M Sohel Rahman

doi:10.1016/j.compbiomed.2022.105385

Revisiting segmentation of lung tumors from CT images

Comput Biol Med. 2022 May:144:105385. doi: 10.1016/j.compbiomed.2022.105385. Epub 2022 Mar 7.

Authors

Farhanaz Farheen¹, Md Salman Shamil¹, Nabil Ibtehaz², M Sohel Rahman³

Affiliations

¹ Department of CSE, BUET, ECE Building, West Palashi, Dhaka, 1230, Bangladesh; Department of CSE, United International University, Dhaka, Bangladesh.
² Department of Computer Science, Purdue University, West Lafayette, IN, United States. Electronic address: nibtehaz@purdue.edu.
³ Department of CSE, BUET, ECE Building, West Palashi, Dhaka, 1230, Bangladesh. Electronic address: msrahman@cse.buet.ac.bd.

PMID: 35299044
DOI: 10.1016/j.compbiomed.2022.105385

Abstract

Lung cancer is a leading cause of death throughout the world. Because the prompt diagnosis of tumors allows oncologists to discern their nature, type, and mode of treatment, tumor detection and segmentation from CT scan images is a crucial field of study. This paper investigates lung tumor segmentation via a two-dimensional Discrete Wavelet Transform (DWT) on the LOTUS dataset (31,247 training, and 4458 testing samples) and a Deeply Supervised MultiResUNet model. Coupling the DWT, which is used to achieve a more meticulous textural analysis while integrating information from neighboring CT slices, with the deep supervision of the model architecture results in an improved dice coefficient of 0.8472. A key characteristic of our approach is its avoidance of 3D kernels (despite being used for a 3D segmentation task), thereby making it quite lightweight.

Keywords: CT scan images; Deep learning; Discrete wavelet transform; Lung tumor; MultiResUNet.

MeSH terms

Humans
Image Processing, Computer-Assisted* / methods
Lung Neoplasms* / diagnostic imaging
Tomography, X-Ray Computed
Wavelet Analysis