Emission textural features I-131 of differentiated thyroid cancer tissue

• Published in: Digital diagnostics Vol. 5; no. 1S; pp. 62 - 64
• Main Authors: Maltsev, Mikhail S., Trukhin, Alexey A., Manaev, Almaz V., Reinberg, Maria V.
• Format: Journal Article
• Language: English
• Published: Eco-Vector 03.07.2024
• Subjects: emission textural features; i-131; thyroid cancer
• ISSN: 2712-8490; 2712-8962
• DOI: 10.17816/DD626496

BACKGROUND: The management of differentiated thyroid cancer includes single-photon emission tomography combined with X-ray computed tomography after radioiodine therapy. Despite a good response to surgery and radioiodine therapy, recurrence is noted in some cases, leading to an unfavorable prognosis in 8% of cases [1]. A preliminary analysis of the distribution of I-131 in residual thyroid tissues and foci of metastasis allows for the estimation of the probability of differentiated cancer recurrence. Currently, there is no method that is simultaneously effective and easy to perform for predicting the recurrence of differentiated thyroid cancer. AIM: The aim of the study was to develop a technique for extracting and computing textural features of the I-131 accumulation region using a single-photon emission tomography system corresponding to differentiated thyroid cancer tissue. MATERIALS AND METHODS: A retrospective analysis of single-photon emission tomography combined with X-ray computed tomography of the neck and thorax of 23 patients was conducted. Regions of interest, including foci of I-131 accumulation in the primary tumor bed, regional and distant metastases, were delineated in Xeleris 4DR software. The obtained mask with the original image was processed in a program written with the help of the Matlab package, which localizes the foci. The textural features of foci are calculated based on the obtained spatial adjacency matrix. This matrix shows how often pixels with certain gray scale brightness values occur in an image. Therefore, the features based on the spatial adjacency matrix reflect the frequency distribution of different pixel neighborhoods in a given context. RESULTS: An algorithm for constructing three-dimensional matrices of a radiation source surrounded by tissue of differentiated thyroid cancer was developed. The textural features of three-dimensional matrices were investigated. It was demonstrated that there are tendencies for differences in texture features corresponding to the ordering of pixel values and image contrast. The values of the obtained features obey the lognormal distribution. CONCLUSIONS: An algorithm for extracting textural features of I-131 accumulation foci allows post-therapy single-photon emission tomography images combined with X-ray computed tomography to be analyzed for the likelihood of recurrence of differentiated thyroid cancer.