Computer aided automatic detection of malignant lesions in diffuse optical mammography

Area of Science:

• Biomedical Optics
• Medical Imaging
• Cancer Diagnostics

Background:

• Diffuse optical tomography (DOT) offers a non-invasive method for imaging breast tissue.
• Current analysis schemes for optical mammography often lack comprehensive statistical approaches.
• Identifying malignancy in breast tissue requires accurate differentiation between healthy and diseased states.

Purpose of the Study:

• To introduce and apply computer-aided detection (CAD) data analysis procedures for deriving composite diffuse optical tomography (DOT) signatures of malignancy in human breast tissue.
• To develop a novel statistical approach utilizing optical property distributions across multiple subjects and voxels.
• To test the methodology in a population of biopsy-confirmed malignant lesions.

Main Methods:

• Employed multiparameter, multivoxel, and multisubject DOT measurements to create a probability of malignancy tomogram.
• Incorporated intrasubject spatial heterogeneity and intersubject physiological property distributions from a training set of cancer-containing breasts.
• Optimized a malignancy parameter (M) using logistic regression to differentiate cancer from healthy voxels, validated on a test set.

Main Results:

• The automated CAD technique successfully generated tomograms distinguishing healthy from malignant breast tissue.
• Achieved an average true positive rate (sensitivity) of 89% and a true negative rate (specificity) of 94% when compared to a gold standard.
• Demonstrated the capability of the developed malignancy parameter (M) in differentiating tissue types.

Conclusions:

• The automated, multisubject, multivoxel, multiparameter statistical analysis of diffuse optical data shows significant potential for distinguishing malignant from healthy breast tissue.
• This approach may enhance the accuracy of breast cancer diagnostics through improved tomogram generation.
• The data analysis methodology could also be beneficial for suppressing image artifacts in DOT.