GPU-accelerated Monte Carlo simulation of MV-CBCT

Is deep learning-enabled real-time personalized CT dosimetry feasible using only patient images as input?

PURPOSE

To propose a novel deep-learning based dosimetry method that allows quick and accurate estimation of organ doses for individual patients, using only their computed tomography (CT) images as input.

METHODS

Despite recent advances in medical dosimetry, personalized CT dosimetry remains a labour-intensive process. Current state-of-the-art methods utilize time-consuming Monte Carlo (MC) based simulations for individual organ dose estimation in CT. The proposed method uses conditional generative adversarial networks (cGANs) to substitute MC simulations with fast dose image generation, based on image-to-image translation. The pix2pix architecture in conjunction with a regression model was utilized for the generation of the synthetic dose images. The lungs, heart, breast, bone and skin were manually segmented to estimate and compare organ doses calculated using both the original and synthetic dose images, respectively.

RESULTS

The average organ dose estimation error for the proposed method was 8.3% and did not exceed 20% for any of the organs considered. The performance of the method in the clinical environment was also assessed. Using segmentation tools developed in-house, an automatic organ dose calculation pipeline was set up. Calculation of organ doses for heart and lung for each CT slice took about 2 s.

CONCLUSIONS

This work shows that deep learning-enabled personalized CT dosimetry is feasible in real-time, using only patient CT images as input.

Impact of a novel multilayer imager on metal artifacts in MV-CBCT

Objective. Megavoltage cone-beam computed tomography (MV-CBCT) imaging offers several advantages including reduced metal artifacts and accurate electron density mapping for adaptive or emergent situations. However, MV-CBCT imaging is limited by the poor efficiency of current detectors. Here we examine a new MV imager and compare CBCT reconstructions under clinically relevant scenarios.Approach. A multilayer imager (MLI), consisting of four vertically stacked standard flat-panel imagers, was mounted to a clinical linear accelerator. A custom anthropomorphic pelvis phantom with replaceable femoral heads was imaged using MV-CBCT and kilovoltage CBCT (kV-CBCT). Bone, aluminum, and titanium were used as femoral head inserts. 8 MU 2.5 MV scans were acquired for all four layers and (as reference) the top layer. Prostate and bladder were contoured on a reference CT and transferred to the other scans after rigid registration, from which the structural similarity index measure (SSIM) was calculated. Prostate and bladder were also contoured on CBCT scans without guidance, and Dice coefficients were compared to CT contours.Main results. kV-CBCT demonstrated the highest SSIMs with bone inserts (prostate: 0.86, bladder: 0.94) and lowest with titanium inserts (0.32, 0.37). Four-layer MV-CBCT SSIMs were preserved with bone (0.75, 0.80) as compared to titanium (0.67, 0.74), outperforming kV-CBCT when metal is present. One-layer MV-CBCT consistently underperformed four-layer results across all phantom configurations. Unilateral titanium inserts and bilateral aluminum insert results fell between the bone and bilateral titanium results. Dice coefficients trended similarly, with four-layer MV-CBCT reducing metal artifact impact relative to KV-CBCT to provide better soft-tissue identification.Significance. MV-CBCT with a four-layer MLI showed improvement over single-layer MV scans, approaching kV-CBCT quality for soft-tissue contrast. In the presence of artifact-producing metal implants, four-layer MV-CBCT scans outperformed kV-CBCT by eliminating artifacts and single-layer MV-CBCT by reducing noise. MV-CBCT with a novel multi-layer imager may be a valuable alternative to kV-CBCT, particularly in the presence of metal.