Radiomics and Quantitative MDA Criteria in Breast Cancer with Bone Metastases by MRI: Examples of Calculation...

Area of Science:

• Oncology
• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Individualized cancer treatment requires rapid and detailed diagnostics.
• Radiomics and AI offer decision support for physicians through quantitative and qualitative assessments.
• Assessing metastatic spinal lesions in breast cancer necessitates advanced analytical tools.

Purpose of the Study:

• To improve quantitative and qualitative assessment of metastatic spinal lesions in breast cancer.
• To enhance the evaluation of treatment-induced changes in spinal metastases.
• To complement existing assessment methods like MDA (Metastasis Detection Algorithm).

Main Methods:

• Utilized MRI data in sagittal projection for a breast cancer patient (T2N3M1).
• Employed machine analysis with image internal structure extraction operators and neural networks.
• Performed analysis of metastatically changed vertebrae structure using machine operators and neural networks.

Main Results:

• Detected subtle structural changes in vertebrae, including "calderas" and altered image complexity patterns during CDK 4/6 inhibitor therapy.
• Neural network-based metastasis recognition increased the reliability of quantitative estimates.
• Demonstrated the ability to record response to therapy and assess its degree compared to previous treatments.

Conclusions:

• Image structure analysis algorithms showed high efficiency in assessing spinal metastases.
• Results correlated well with radiologist's opinion and clinical/laboratory data.
• Enabled analysis of subtle effects, yielding new qualitative insights beyond quantitative metrics.