Improve follicular thyroid carcinoma diagnosis using computer aided diagnosis system on ultrasound images

Efficacy of color Doppler ultrasound signs combined with serum tumor-specific growth factor in the diagnosis of differentiated thyroid cancer

OBJECTIVE

To construct a diagnostic model for follicular thyroid carcinoma (FTC) and papillary thyroid carcinoma (PTC), both subtypes of differentiated thyroid carcinoma (DTC), using color Doppler ultrasound signs in conjunction with serum laboratory markers.

METHODS

We conducted a retrospective analysis of patients with thyroid nodules who underwent ultrasonography at Yulin Hospital from February 2021 to March 2023. The cohort included 269 subjects: 105 with benign nodules and 164 with DTC (59 with FTC and 105 with PTC). We compared baseline demographics and laboratory indices between the groups. Diagnostic values of ultrasound features and laboratory markers were assessed using receiver operating characteristic (ROC) curves, and logistic regression was employed to pinpoint independent diagnostic factors for FTC. A predictive nomogram was subsequently developed based on these factors.

RESULTS

There were significant differences between the benign and malignant groups regarding ultrasound signs (including border, morphology, echogenicity, calcification, blood flow, lymph node zoning) and laboratory indices (free triiodothyronine (FT3), free thyroxine (FT4), thyroglobulin (Tg), thyroid-stimulating hormone (TSH), vascular endothelial growth factor (VEGF), tumor-specific growth factor (TSGF)), with all P-values <0.05. The areas under the curve (AUCs) for FT3, FT4, Tg, TSH, VEGF, and TSGF were all above 0.75, with Tg achieving the highest at 0.91. Logistic regression identified borders, morphology, echogenicity, VEGF, and TSGF as independent diagnostic factors for distinguishing between FTC and PTC, with significant P-values. The constructed nomogram demonstrated an AUC of 0.853, indicating high diagnostic accuracy. Both calibration and decision curve analysis (DCA) validated the model's stability and clinical utility.

CONCLUSION

We successfully developed a nomogram combining ultrasound features and serum markers that enhances the diagnostic precision for FTC. This model offers a valuable tool for clinical diagnostics in differentiated thyroid cancer.

Long Non-Coding RNAs as Determinants of Thyroid Cancer Phenotypes: Investigating Differential Gene Expression Patterns and Novel Biomarker Discovery

Thyroid Cancer (TC) is the most common endocrine malignancy, with increasing incidence globally. Papillary thyroid cancer (PTC), a differentiated form of TC, accounts for approximately 90% of TC and occurs predominantly in women of childbearing age. Although responsive to current treatments, recurrence of PTC by middle age is common and is much more refractive to treatment. Undifferentiated TC, particularly anaplastic thyroid cancer (ATC), is the most aggressive TC subtype, characterized by it being resistant and unresponsive to all therapeutic and surgical interventions. Further, ATC is one of the most aggressive and lethal malignancies across all cancer types. Despite the differences in therapeutic needs in differentiated vs. undifferentiated TC subtypes, there is a critical unmet need for the identification of molecular biomarkers that can aid in early diagnosis, prognosis, and actionable therapeutic targets for intervention. Advances in the field of cancer genomics have enabled for the elucidation of differential gene expression patterns between tumors and healthy tissue. A novel category of molecules, known as non-coding RNAs, can themselves be differentially expressed, and extensively contribute to the up- and downregulation of protein coding genes, serving as master orchestrators of regulated and dysregulated gene expression patterns. These non-coding RNAs have been identified for their roles in driving carcinogenic patterns at various stages of tumor development and have become attractive targets for study. The identification of specific genes that are differentially expressed can give insight into mechanisms that drive carcinogenic patterns, filling the gaps of deciphering molecular and cellular processes that modulate TC subtypes, outside of well-known driver mutations.

Diagnostic value of deep learning-assisted endoscopic ultrasound for pancreatic tumors: a systematic review and meta-analysis

Background and aims

Endoscopic ultrasonography (EUS) is commonly utilized in the diagnosis of pancreatic tumors, although as this modality relies primarily on the practitioner's visual judgment, it is prone to result in a missed diagnosis or misdiagnosis due to inexperience, fatigue, or distraction. Deep learning (DL) techniques, which can be used to automatically extract detailed imaging features from images, have been increasingly beneficial in the field of medical image-based assisted diagnosis. The present systematic review included a meta-analysis aimed at evaluating the accuracy of DL-assisted EUS for the diagnosis of pancreatic tumors diagnosis.

Methods

We performed a comprehensive search for all studies relevant to EUS and DL in the following four databases, from their inception through February 2023: PubMed, Embase, Web of Science, and the Cochrane Library. Target studies were strictly screened based on specific inclusion and exclusion criteria, after which we performed a meta-analysis using Stata 16.0 to assess the diagnostic ability of DL and compare it with that of EUS practitioners. Any sources of heterogeneity were explored using subgroup and meta-regression analyses.

Results

A total of 10 studies, involving 3,529 patients and 34,773 training images, were included in the present meta-analysis. The pooled sensitivity was 93% (95% confidence interval [CI], 87-96%), the pooled specificity was 95% (95% CI, 89-98%), and the area under the summary receiver operating characteristic curve (AUC) was 0.98 (95% CI, 0.96-0.99).

Conclusion

DL-assisted EUS has a high accuracy and clinical applicability for diagnosing pancreatic tumors.

Systematic review registration

https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023391853, identifier CRD42023391853.