Computer aided prognosis for cell death categorization and prediction in vivo using quantitative ultrasound and
M J Gangeh1, A Hashim2, A Giles2
1Departments of Medical Biophysics, and Radiation Oncology, University of Toronto, Toronto, Ontario M5G 2M9, Canada and Departments of Radiation Oncology, and Imaging Research - Physical Sciences, Sunnybrook Health Sciences Centre, Toronto, Ontario M4N 3M5, Canada.
Medical Physics
|December 3, 2016
Summary
A new computer-aided-prognosis (CAP) system uses quantitative ultrasound (QUS) and machine learning to noninvasively predict cancer treatment response. This tool helps assess cell death levels, enabling personalized cancer therapy and early detection of treatment resistance.
Area of Science:
- Medical Imaging
- Oncology
- Machine Learning
Background:
- Current cancer therapy lacks imaging-based standards for monitoring treatment response.
- A one-size-fits-all approach is often used, highlighting the need for personalized treatment assessment.
Purpose of the Study:
- To develop a computer-aided-prognosis (CAP) system for noninvasive assessment of cancer treatment response.
- To categorize and predict cell death levels in fibrosarcoma tumors using quantitative ultrasound (QUS) spectroscopy, texture analysis, and machine learning.
Main Methods:
- Ultrasound-stimulated microbubbles and x-ray radiation therapy were used to treat sarcoma xenograft tumors in mice.
- High-frequency ultrasound assessed treatment effects, generating 2D spectral parametric maps.
- Machine learning classifiers and regressors predicted cell death levels based on textural features from QUS data.
Main Results:
- The CAP system achieved high accuracy in classifying cell death levels (AUC=0.87).
- Predicted cell death levels showed good correlation with histological data (r=0.68, p<0.001).
- The system identified significant changes in cell death, mirroring ground-truth findings.
Conclusions:
- The developed CAP system provides a noninvasive framework for quantifying cancer treatment response.
- This technology addresses a critical gap in cancer care, enabling enhanced treatment decision-making.
- It has the potential to facilitate early detection of treatment resistance, allowing for timely therapy adjustments.
