Utilizing DETR model on SPECT image to assess remaining thyroid tissues post-thyroidectomy

In the management of thyroid cancer, subsequent to the excision of thyroid tissue through surgery, patients are administered with I-131 medication to eliminate residual thyroid tissue. An integral stage of the treatment process is the assessment of successful tissue removal. In this research, we present a model that aids in the automatic evaluation of a patient’s tissue removal using SPECT thyroid scan images. In instances where there is unsuccessful removal of tissue, the model is capable of automatically identifying residual thyroid tissue. For diagnostic purposes, the RSI is employed, and the detection model utilized is DETR, which effectively resolves the challenge of distance learning, thereby leading to desirable outcomes. Specifically, our proposed model achieves 0.91 on F1-score, mAP@0.5 with thyroid reaching 0.73, thereby attesting to its effectiveness. Therefore, our proposal is expected to contribute to the development of medical technology and improve the quality of diagnosis of thyroid nodule removal.