Neural network-assisted automated image registration for MRI-guided adaptive brachytherapy in cervical cancer

PURPOSE

In image-guided adaptive brachytherapy (IGABT) a quantitative evaluation of the dosimetric changes between fractions due to anatomical variations, can be implemented via rigid registration of images from subsequent fractions based on the applicator as a reference structure. With available treatment planning systems (TPS), this is a manual and time-consuming process. The aim of this retrospective study was to automate this process. A neural network (NN) was trained to predict the applicator structure from MR images. The resulting segmentation was used to automatically register MR-volumes.

MATERIAL AND METHODS

DICOM images and plans of 56 patients treated for cervical cancer with high dose-rate (HDR) brachytherapy were used in the study. A 2D and a 3D NN were trained to segment applicator structures on clinical T2-weighted MRI datasets. Different rigid registration algorithms were investigated and compared. To evaluate a fully automatic registration workflow, the NN-predicted applicator segmentations (AS) were used for rigid image registration with the best performing algorithm. The DICE coefficient and mean distance error between dwell positions (MDE) were used to evaluate segmentation and registration performance.

RESULTS

The mean DICE coefficient for the predicted AS was 0.70 ± 0.07 and 0.58 ± 0.04 for the 3D NN and 2D NN, respectively. Registration algorithms achieved MDE errors from 8.1 ± 3.7 mm (worst) to 0.7 ± 0.5 mm (best), using ground-truth AS. Using the predicted AS from the 3D NN together with the best registration algorithm, an MDE of 2.7 ± 1.4 mm was achieved.

CONCLUSION

Using a combination of deep learning models and state of the art image registration techniques has been demonstrated to be a promising solution for automatic image registration in IGABT. In combination with auto-contouring of organs at risk, the auto-registration workflow from this study could become part of an online-dosimetric interfraction evaluation workflow in the future.

Auto-segmentation technique for SEM images using machine learning: Asphaltene deposition case study

Enormous efforts have been attempted to replace the subjective human analysis of digital images by automated computational methods such as image processing and computer vision. The image processing methods (i.e., pixel- and object-based assessments) determine a variety of spectral/optical features and devices/data properties on the images so that appropriate information could be extracted. However, these methods may not be appropriate when globally heterogeneous and locally anisotropic features exist such as those found in Scanning Electron Microscopy (SEM) Images. Thus, it is essential to have an adaptive and data-driven procedure to extract optimal information from individual SEM images. In this study, we developed a fully automated image processing and analysis method using data analytics, pattern recognition, and machine learning (including deep learning) techniques to automate the image processing and investigate physical properties and nanoscale deposition of petroleum constituents such as asphaltenes on surfaces from over hundreds of images. To do so, various data preparation processes (i.e., image filtering and quality assessment methods) were first introduced to mine and enhance the data. Then, the extracted information was used to identify and quantify targeted physical properties and deposition attributes by denosing and image segmentation techniques. To validate the proposed method, we applied the model to the experimental results from asphaltene deposition studies. The model results were then compared with the corresponding experimental counterparts from the literature. The insight from this application led to a better understanding of the asphaltene deposition mechanism. To the best of our knowledge, this work is one of the first attempts to develop a fully automated image-processing model that describes physical properties of deposited species. In addition, our combination of data and descriptive models allowed us to differentiate foreground and background information particles from SEM images meaning that the model could be used for other applications in image processing.