Search

Significance: The combination of polarized imaging with artificial intelligence (AI) technology has provided a powerful tool for performing an objective and precise diagnosis in medicine., Aim: An approach is proposed for the detection of hepatitis B (HB) virus using a combined Mueller matrix imaging technique and deep learning method., Approach: In the proposed approach, Mueller matrix imaging polarimetry is applied to obtain 4 × 4 Mueller matrix images of 138 HBsAg-containing (positive) serum samples and 136 HBsAg-free (negative) serum samples. The kernel estimation density results show that, of the 16 Mueller matrix elements, elements M 22 and M 33 provide the best discriminatory power between the positive and negative samples., Results: As a result, M 22 and M 33 are taken as the inputs to five different deep learning models: Xception, VGG16, VGG19, ResNet 50, and ResNet150. It is shown that the optimal classification accuracy (94.5%) is obtained using the VGG19 model with element M 22 as the input., Conclusions: Overall, the results confirm that the proposed hybrid Mueller matrix imaging and AI framework provides a simple and effective approach for HB virus detection., (© 2022 The Authors.)

Detection of ultrasound (US)-modulated fluorescence in turbid media is a challenge because of the low level of fluorescent light and the weak modulation of incoherent light. A very limited number of theoretical and experimental investigations have been performed, and this is, to our knowledge, the first demonstration of pulsed US-modulated fluorescence tomography. Experimental results show that the detected signal depends on the acoustic frequency and the fluorescent target's size along the ultrasonic propagation axis. The modulation depth of the detected signal is greatest when the length of the object along the acoustic axis is an odd number of half wavelengths and is weakest when the object is an integer multiple of an acoustic wavelength. Images of a fluorescent tube embedded within a 22- by 13- by 30 mm scattering gel phantom (μ(s)∼15  cm(-1), g=0.93) with 1-, 1.5-, and 2 MHz frequency US are presented. The modulation depth of the detected signal changes by a factor of 5 depending on the relative size of the object and the frequency. The approach is also verified by some simple experiments in a nonscattering gel and using a theoretical model.

In pulsed ultrasound modulated optical tomography (USMOT), an ultrasound (US) pulse performs as a scanning probe within the sample as it propagates, modulating the scattered light spatially distributed along its propagation axis. Detecting and processing the modulated signal can provide a 1-dimensional image along the US axis. A simple model is developed wherein the detected signal is modelled as a convolution of the US pulse and the properties (ultrasonic/optical) of the medium along the US axis. Based upon this model, a maximum likelihood (ML) method for image reconstruction is established. For the first time to our knowledge, the ML technique for an USMOT signal is investigated both theoretically and experimentally. The ML method inverts the data to retrieve the spatially varying properties of the sample along the US axis, and a signal proportional to the optical properties can be acquired. Simulated results show that the ML method can serve as a useful reconstruction tool for a pulsed USMOT signal even when the signal-to-noise ratio (SNR) is close to unity. Experimental data using 5 cm thick tissue phantoms (scattering coefficient μ(s) = 6.5 cm(-1), anisotropy factor g=0.93) demonstrate that the axial resolution is 160 μm and the lateral resolution is 600 μm using a 10 MHz transducer.