Weakly supervised learning-based 3D bladder reconstruction from 2D ultrasound images for bladder volume measurement.

BACKGROUND: Accurate measurement of bladder volume is necessary to maintain the consistency of the patient's anatomy in radiation therapy for pelvic tumors. As the diversity of the bladder shape, traditional methods for bladder volume measurement from 2D ultrasound have been found to produce inaccurate results. PURPOSE: To improve the accuracy of bladder volume measurement from 2D ultrasound images for patients with pelvic tumors. METHODS: The bladder ultrasound images from 130 patients with pelvic cancer were collected retrospectively. All data were split into a training set (80 patients), a validation set (20 patients), and a test set (30 patients). A total of 12 transabdominal ultrasound images for one patient were captured by automatically rotating the ultrasonic probe with an angle step of 15°. An incomplete 3D ultrasound volume was synthesized by arranging these 2D ultrasound images in 3D space according to the acquisition angles. With this as input, a weakly supervised learning-based 3D bladder reconstruction neural network model was built to predict the complete 3D bladder. The key point is that we designed a novel loss function, including the supervised loss of bladder segmentation in the ultrasound images at known angles and the compactness loss of the 3D bladder. Bladder volume was calculated by counting the number of voxels belonging to the 3D bladder. The dice similarity coefficient (DSC) was used to evaluate the accuracy of bladder segmentation, and the relative standard deviation (RSD) was used to evaluate the calculation accuracy of bladder volume with that of computed tomography (CT) images as the gold standard. RESULTS: The results showed that the mean DSC was up to 0.94 and the mean absolute RSD can be reduced to 6.3% when using 12 ultrasound images of one patient. Further, the mean DSC also was up to 0.90 and the mean absolute RSD can be reduced to 9.0% even if only two ultrasound images were used (i.e., the angle step is 90°). Compared with the commercial algorithm in bladder scanners, which has a mean absolute RSD of 13.6%, our proposed method showed a considerably huge improvement. CONCLUSIONS: The proposed weakly supervised learning-based 3D bladder reconstruction method can greatly improve the accuracy of bladder volume measurement. It has great potential to be used in bladder volume measurement devices in the future.

Deep learning based automatic internal gross target volume delineation from 4D-CT of hepatocellular carcinoma patients.

BACKGROUND: The location and morphology of the liver are significantly affected by respiratory motion. Therefore, delineating the gross target volume (GTV) based on 4D medical images is more accurate than regular 3D-CT with contrast. However, the 4D method is also more time-consuming and laborious. This study proposes a deep learning (DL) framework based on 4D-CT that can achieve automatic delineation of internal GTV. METHODS: The proposed network consists of two encoding paths, one for feature extraction of adjacent slices (spatial slices) in a specific 3D-CT sequence, and one for feature extraction of slices at the same location in three adjacent phase 3D-CT sequences (temporal slices), a feature fusion module based on an attention mechanism was proposed for fusing the temporal and spatial features. Twenty-six patients' 4D-CT, each consisting of 10 respiratory phases, were used as the dataset. The Hausdorff distance (HD95), Dice similarity coefficient (DSC), and volume difference (VD) between the manual and predicted tumor contour were computed to evaluate the model's segmentation accuracy. RESULTS: The predicted GTVs and IGTVs were compared quantitatively and visually with the ground truth. For the test dataset, the proposed method achieved a mean DSC of 0.869 ± 0.089 and an HD95 of 5.14 ± 3.34 mm for all GTVs, with under-segmented GTVs on some CT slices being compensated by GTVs on other slices, resulting in better agreement between the predicted IGTVs and the ground truth, with a mean DSC of 0.882 ± 0.085 and an HD95 of 4.88 ± 2.84 mm. The best GTV results were generally observed at the end-inspiration stage. CONCLUSIONS: Our proposed DL framework for tumor segmentation on 4D-CT datasets shows promise for fully automated delineation in the future. The promising results of this work provide impetus for its integration into the 4DCT treatment planning workflow to improve hepatocellular carcinoma radiotherapy.

Uncertainty-guided man-machine integrated patient-specific quality assurance.

PURPOSE: Providing the confidence level (Uncertainty) of prediction results and guiding patient-specific quality assurance (pQA) can enhance the safety of AI (Artificial intelligence)-based automatic pQA models. However, even state-of-the-art automatic pQA models can only predict the gamma passing rate (GPR) and cannot quantify the prediction uncertainty, limiting the safe clinical translation of these models. This study aims to develop an uncertainty-guided man-machine integrated pQA (UgMi-pQA) method to address this issue. METHODS: An uncertainty-aware dual-task deep learning (UDDL) model, combined with an interwoven training method and Monte Carlo dropout approximation Bayesian inference, to enable simultaneous output of the predicted GPR and corresponding total prediction uncertainty to guide pQA. 1541 pairs of field fluences and GPRs collected from 165 glioma, 50 lung (conventional fractionation), and 20 liver cases were separated for the UDDL model training, validation, calibration, and test in a ratio of 7:1:1:1, respectively. Furthermore, 413 pairs of fluences and GPRs collected from 12 breast, 10 cervix, 9 esophagus, 8 tongue, and 12 lung SBRT cases were gathered for the out-of-distribution (OOD) detection. RESULTS: Clinical accuracy of 100.0% was reached with only 61.7% of the workload. Samples with substantial prediction errors and failed samples with low label GPR (<95%) could be successfully screened out. The capability ranges of two different models were both successfully identified with the prediction uncertainty significantly larger for OOD samples than for in-distribution samples (p < 0.01). CONCLUSION: This study presents the first work on uncertainty quantification for deep learning automatic pQA tasks. The UgMi-pQA method can balance the efficiency and safety of the automatic pQA models and promote their clinical application.

Improving the accuracy of converting dose to medium to dose to water algorithms in small megavoltage photon fields in dose to medium based treatment planning systems.

PURPOSE: To present a formalism to improve the accuracy of converting absorbed dose to medium in medium (Dm) to absorbed dose to water in medium (Dw) in small megavoltage photon fields for different human tissues in Dm-based treatment planning systems (TPS). METHODS: Eight kinds of real human tissues were simulated to convert Dm to Dw. Four kinds of virtual water media were deliberately designed to analyze source of deviations from the conventional Bragg-Gray theory. Mass electronic stopping powers were calculated using the ESTAR code. The phase-space data was generated by the EGSnrc/BEAMnrc Monte Carlo code. The dose deposition was calculated with the EGSnrc/DOSRZnrc code. Electron fluence spectra calculated with EGSnrc/FLURZnrc code were utilized to analyze fluence perturbations and determine fluence intensity (Φw,mint) and fluence spectral shape (Φw,mS) correction factors. RESULTS: Large conversion errors of Dw using Bragg-Gray theory were observed, such as 19.65% ± 9.58% (average value ± standard deviation, type A) for inflated lung (ICRU). Fluence perturbations could be exacerbated by severe charged particle disequilibrium conditions. These deviations were caused by the synergy between tissues' different mean excitation energies and smaller mass densities compared to those of water. Adding Φw,mint and Φw,mS correction factors to modify Bragg-Gray theory could greatly reduce Dw conversion errors, within 1.00% for all tissues studied. CONCLUSIONS: The current clinically used Dw conversion algorithm in commercial Dm-based TPS isn't appropriate for some human tissues in small field dosimetry. Correction factors should be exploited to improve the accuracy.

A novel crystal-analyzer phase retrieval algorithm and its noise property.

A description of the rocking curve in diffraction enhanced imaging (DEI) is presented in terms of the angular signal response function and a simple multi-information retrieval algorithm based on the cosine function fitting. A comprehensive analysis of noise properties of DEI is also given considering the noise transfer characteristic of the X-ray source. The validation has been performed with synchrotron radiation experimental data and Monte Carlo simulations based on the Geant4 toolkit combined with the refractive process of X-rays, which show good agreement with each other. Moreover, results indicate that the signal-to-noise ratios of the refraction and scattering images are about one order of magnitude better than that of the absorption image at the edges of low-Z samples. The noise penalty is drastically reduced with the increasing photon flux and visibility. Finally, this work demonstrates that the analytical method can build an interesting connection between DEI and GDPCI (grating-based differential phase contrast imaging) and is widely suitable for a variety of measurement noise in the angular signal response imaging prototype. The analysis significantly contributes to the understanding of noise characteristics of DEI images and may allow improvements to the signal-to-noise ratio in biomedical and material science imaging.

High sensitivity phase retrieval method in grating-based x-ray phase contrast imaging.

PURPOSE: Grating-based x-ray phase contrast imaging is considered as one of the most promising techniques for future medical imaging. Many different methods have been developed to retrieve phase signal, among which the phase stepping (PS) method is widely used. However, further practical implementations are hindered, due to its complex scanning mode and high radiation dose. In contrast, the reverse projection (RP) method is a novel fast and low dose extraction approach. In this contribution, the authors present a quantitative analysis of the noise properties of the refraction signals retrieved by the two methods and compare their sensitivities. METHODS: Using the error propagation formula, the authors analyze theoretically the signal-to-noise ratios (SNRs) of the refraction images retrieved by the two methods. Then, the sensitivities of the two extraction methods are compared under an identical exposure dose. Numerical experiments are performed to validate the theoretical results and provide some quantitative insight. RESULTS: The SNRs of the two methods are both dependent on the system parameters, but in different ways. Comparison between their sensitivities reveals that for the refraction signal, the RP method possesses a higher sensitivity, especially in the case of high visibility and/or at the edge of the object. CONCLUSIONS: Compared with the PS method, the RP method has a superior sensitivity and provides refraction images with a higher SNR. Therefore, one can obtain highly sensitive refraction images in grating-based phase contrast imaging. This is very important for future preclinical and clinical implementations.

Three-dimensional study of poly(lactic co-glycolic acid) micro-porous microspheres using hard X-ray nano-tomography.

Poly(lactic co-glycolic acid) (PLGA) is widely used in diverse fields, especially in delivering biologically active proteins and drugs. For these applications, the knowledge of morphology and microstructure of PLGA micro-porous microspheres is of great importance since they strongly influence the drug delivering efficiency. In this study, micro-porous PLGA microspheres loaded by bovine serum albumin are investigated by using a full-field Zernike phase contrast transmission hard X-ray microscope. From three-dimensional reconstructions and segmentations, fundamental microstructural parameters such as size, shape, distribution and volume ratio among pores and proteins inside PLGA microspheres were obtained. These parameters are useful to understand the relationship between the internal microstructure and drug encapsulation, as well as the drug release efficiency of PLGA microspheres. The presented results demonstrate the capability of hard X-ray nano-tomography to characterize porous microspheres loaded with proteins and drugs, and also open a way to analyse, optimize and design new PLGA microspheres for specific applications.