A Systematic Review of the Techniques for the Automatic Segmentation of Organs-at-Risk in Thoracic Computed Tomography Images

• Published in: Archives of computational methods in engineering Vol. 28; no. 4; pp. 3245 - 3267
• Main Authors: Ashok, Malvika, Gupta, Abhishek
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.06.2021; Springer Nature B.V
• Subjects: Accuracy; Aorta; Computed tomography; Engineering; Esophagus; Image segmentation; Mathematical and Computational Engineering; Medical imaging; Metric space; Organs; Original Paper; Performance evaluation; Questions; Radiation therapy; Risk; Systematic review; Tomography; Trachea
• ISSN: 1134-3060; 1886-1784
• DOI: 10.1007/s11831-020-09497-z

The standard treatment for the cancer is the radiotherapy where the organs nearby the target volumes get affected during treatment called the Organs-at-risk. Segmentation of Organs-at-risk is crucial but important for the proper planning of radiotherapy treatment. Manual segmentation is time consuming and tedious in regular practices and results may vary from experts to experts. The automatic segmentation will produce robust results with precise accuracy. The aim of this systematic review is to study various techniques for the automatic segmentation of organs-at-risk in thoracic computed tomography images and to discuss the best technique which give the higher accuracy in terms of segmentation among all other techniques proposed in the literature. PRISMA guidelines had been used to conduct this systematic review. Three online databases had been used for the identification of the related papers and a query had been formed for the search purpose. The papers were shortlisted based on the various inclusion and exclusion criteria. Four research questions had been designed and answers of those were explored. After reviewing all the techniques, the best technique had been selected and discussed in detail which gave the precise accuracy based on Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD). Both DSC and HD were used in the literature to evaluate the performance of their proposed technique for the automatic segmentation of four organs (esophagus, heart, trachea and aorta). However, the value of these parameters vary as per the validation sample size. Consequently, various challenges faced by the researchers had been listed. This paper includes the summary of the various automatic segmentation techniques for the Organs-at-risk in thoracic computed tomography images in terms of four research questions. Different techniques, Datasets, Performance accuracy and various challenges had been discussed.