Concept Development of an On-Chip PET System.

Organs-on-Chips (OOCs), microdevices mimicking in vivo organs, find growing applications in disease modeling and drug discovery. With the increasing number of uses comes a strong demand for imaging capabilities of OOCs. Positron Emission Tomography (PET) would be ideal for OOC imaging, however, current PET systems have insufficient spatial resolution for this task. In this work, we propose the concept of an On-Chip PET system capable of imaging OOCs. Our system consists of four detectors arranged around the OOC device. Each detector is made of two monolithic Lutetium-yttrium oxyorthosilicate (LYSO) crystals and covered with Silicon photomultipliers (SiPMs) on multiple surfaces. We use a Convolutional Neural Network (CNN) trained with data from a Monte Carlo Simulation (MCS) to predict the first gamma-ray interaction position inside the detector from the light patterns that are recorded by the SiPMs on the detector's surfaces. With the Line of Responses (LORs) created by the predicted interaction positions, we reconstruct with Simultaneous Algebraic Reconstruction Technique (SART). The CNN achieves a mean average prediction error of 0.78 mm in the best configuration. We use the trained network to reconstruct an image of a grid of 21 point sources spread across the field-of-view and obtain a mean spatial resolution of 0.53 mm. We demonstrate that it is possible to achieve a spatial resolution of almost 0.5 mm in a PET system made of multiple monolithic LYSO crystals by directly predicting the scintillation position from light patterns created with SiPMs. We observe that CNNs from the ResNet family perform better than those from the EfficientNet family and that certain surfaces encode significantly more information for the scintillation-point prediction than others.

Deep Learning for Predicting Gamma-Ray Interaction Positions in LYSO Detector.

Positron Emission Tomography (PET) is among the most commonly used medical imaging modalities in clinical practice, especially for oncological applications. In contrast to conventional imaging modalities like X-ray Computed Tomography (CT) or Magnetic Resonance Imaging (MRI), PET retrieves in vivo information about biochemical processes rather than just anatomical structures. However, physical limitations and detector constraints lead to an order of magnitude lower spatial resolution in PET images. In recent years, the use of monolithic detector crystals has been investigated to overcome some of the factors limiting spatial resolution. The key to increasing PET systems' resolution is to estimate the gamma-ray interaction position in the detector as precisely as possible.In this work, we evaluate a Convolutional Neural Network (CNN) based reconstruction algorithm that predicts the gamma-ray interaction position using light patterns recorded with Silicon photomultipliers (SiPMs) on the crystal's surfaces. The algorithm is trained on data from a Monte Carlo Simulation (MCS) that models a gamma point source and a detector consisting of Lutetium-yttrium oxyorthosilicate (LYSO) crystals and SiPMs added to five surfaces. The final Mean Absolute Error (MAE) on the test dataset is 1.48 mm.

Introduction of external magnetic fields in entropic moment modelling for radiotherapy.

One of the big challenges of the emerging MRI-guided radiotherapy is the prediction of an external magnetic field effect on the deposited dose induced by a beam of charged particles. In this paper, we present the results of the implementation of the Lorentz force in the deterministic M1 model. The validation of our code is performed by comparisons with the Monte-Carlo code FLUKA. The relevant examples show a significant modification of the shape of dose deposition volume induced by the external magnetic field in presence of heterogeneities. A gamma-index analysis 3%/3mm shows a good agreement of our model with FLUKA simulations.

Deterministic model for the transport of energetic particles: Application in the electron radiotherapy.

A new deterministic method for calculating the dose distribution in the electron radiotherapy field is presented. The aim of this work was to validate our model by comparing it with the Monte Carlo simulation toolkit, GEANT4. A comparison of the longitudinal and transverse dose deposition profiles and electron distributions in homogeneous water phantoms showed a good accuracy of our model for electron transport, while reducing the calculation time by a factor of 50. Although the Bremsstrahlung effect is not yet implemented in our model, we propose here a method that solves the Boltzmann kinetic equation and provides a viable and efficient alternative to the expensive Monte Carlo modeling.