Intelligent Photoacoustic Diagnosis of Osteoporosis Based on Wavelet Scattering Network

Cancellous bone is a porous medium with fluid bone marrow clusters filling the solid bone trabecular framework. The main challenges of multi-wavelength photoacoustic (PA) analysis for cancellous bone assessment are the high complexity and low signal-to-noise ratio of PA signals due to multiple scattering. To solve these problems, we proposed an intelligent method based on a wavelet scattering network (WSN) for accurately assessing bone quality and diagnosing diseases. Signals from rabbit models of osteoporosis obtained by the homemade PA detection system were processed by our method. The results show that the multi-wavelength scattering features obtained by WSN as the input to the support vector machine can achieve 100% correct intelligent classification of osteoporotic and normal bone samples, indicating WSN can accurately extract the scattering features containing physicochemical information hidden in the multi-wavelength PA signals of bone. In conclusion, with the help of WSN, intelligent diagnosis of bone diseases based on multi-wavelength photoacoustic sensing can be achieved.