Quantification and classification of echographic findings in the thyroid gland by computerized B-mode texture...

Area of Science:

• Medical Imaging
• Diagnostic Ultrasound
• Quantitative Echography

Background:

• Standard thyroid nodule diagnosis relies on conventional ultrasound and needle biopsy.
• Single needle biopsies may not be sufficient to rule out malignancy in all cases.
• There is a need for improved diagnostic accuracy in thyroid echography.

Purpose of the Study:

• To establish quantitative criteria for thyroid echography using computerized B-mode image analysis.
• To assess the diagnostic accuracy of computerized ultrasound in identifying thyroid disease.
• To identify statistical parameters for effective thyroid lesion characterization.

Main Methods:

• Pilot study involving computerized B-mode image analysis of thyroid ultrasound data.
• Evaluation of statistical parameters, including the ten per cent percentile of grey level distribution and entropy.
• Comparison of computerized ultrasound findings with conventional diagnostic methods.

Main Results:

• The ten per cent percentile of grey level distribution and entropy were identified as the most efficient statistical parameters.
• Computerized ultrasound detected all nodular lesions with an overall accuracy of approximately 85%.
• Computer-aided ultrasound effectively identified echographic characteristics indicative of high malignancy risk in thyroid lesions.

Conclusions:

• Computerized ultrasound analysis provides quantitative criteria for thyroid echography.
• The procedure demonstrates significant diagnostic accuracy for thyroid disease, particularly in nodular lesions.
• Computerized analysis can guide decisions for repeat needle biopsies in cases with suspicious findings, even after a negative initial biopsy.