An Automated Preprocessing Method for Diffuse Optical Tomography to Improve Breast Cancer Diagnosis

Area of Science:

• Biomedical Imaging
• Optical Physics
• Oncology

Background:

• Ultrasound-guided diffuse optical tomography (uDOG) is a noninvasive technique for breast cancer diagnosis.
• Accurate estimation of optical absorptions and total hemoglobin concentration is crucial for uDOG.
• Measurement errors can significantly reduce the accuracy of uDOG.

Purpose of the Study:

• To introduce an automated preprocessing method for detecting and correcting outliers in uDOG measurements.
• To improve the accuracy of total hemoglobin concentration estimation in breast lesions.
• To enhance the classification of malignant and benign breast lesions.

Main Methods:

• Developed an automated preprocessing method combining data from 4 optical wavelengths.
• Introduced new measures: correlation between wavelength pairs and wavelength consistency index.
• Applied the method to phantom and patient data, including cases with measurement errors.

Main Results:

• The automated method effectively detected and corrected measurement errors.
• Correlation coefficients improved significantly after applying the correction method.
• Wavelength consistency index normalized for datasets with and without errors.
• Improved classification accuracy for malignant and benign lesions.

Conclusions:

• The automated preprocessing method enhances the reliability of uDOG for breast cancer diagnosis.
• The technique improves the accuracy of total hemoglobin concentration and lesion classification.
• This approach offers a robust solution for mitigating measurement errors in optical imaging.