The major activities of this research project in 2019 have centered around optimizing the methods and procedures for imaging rhesus macaques, NZW rabbits, and common marmosets on new scanners while conducting a variety of chemotherapy and basic immunology experiments in Mtb models. As a part of these efforts we have optimized scanner quality control systems, data control systems, anesthesia protocols, ventilation procedures and breath-hold methods to produce the best images as safely as possible for the animal subjects and the experiments being conducted. The process is still ongoing, but a standard protocol for medium sized (3 to 6 kg) animals was established and has been applied successfully in Mtb-infected rhesus and NZW rabbits. As we have more than 10 years of data and multiple publications using CT Hounsfield unit (HU)density ranges and PET FDG uptake values as descriptive and quantitative features for tuberculosis lesions, we have made systematic study of the quantitative differences in images collected on the previous small clinical CT scanner and the new scanner (LFER). One study used a common CATPHAN phantom that was imaged on both systems. This phantom, typically used for scanner quality control (QC), with six standard materials with varying mean HUs that mimic densities found in a live subject (HUs similar to bone, air, muscle, fat) was imaged on both scanners with several energy settings (kVp and mAs). After plotting the linear attenuation coefficient (LAC) of the varying materials and the HU measurement of the different materials, it was found both scanners have a similar linear slope with differing y intercepts of approximately 115 HU. The data are comparable with the LFER having a HU around 115 HU lower than the clinical scanner. This correlation may be related to the very different CT designs and techniques used to collect the projections. To choose the best possible technique for imaging the rhesus macaque, tests were run with the CATPHAN phantom to check which setting combination reported mean HU ranges similar to the standard published HU range for the materials imaged. After running these tests, it was found that the optimal technique available on the LFER was 80 kVp 980 A 80 ms (65.6 As) for the rhesus and the rabbit. Scans using this technique were collected and analyzed with specific attention to the HU ranges where lesions were identified, thus the density ranges for lesions on the new CT scans is being perfected. Additional parameters that must be examined for collecting an optimal PET image are the probe dose administered to the subjects, the time it is allowed to dwell, and the duration of the data collection. The new system was reported to have greater sensitivity than our previous system and it also had a different bore configuration. With PET detectors, it is important to optimize the concentration and distribution of the radioactivity within the bore so that the detectors function rate is not exceeded nor is the activity in any one location is not too high to be counted nor so low that no signal is received for that voxel location. This meant that we had to determine the ideal FDG dose that would best monitor changes TB lesions in each of the various species. We were able to successfully identify and establish an optimal FDG dose for the rhesus through serial PET/CT scans on several rhesus macaques infected with Mtb using 3 different 18F-2-fluoro-2-deoxy-D-glucose (FDG) doses, 0.2mCI/kg, 0.5mCi/kg, 1.0mCi/kg and 2.0mCi/kg. A thorough dose study showed that 0.5 mCi/Kg was ideal to minimize both noise and variability in the resulting images. To the naked eye, the images resulting from various doses were similar, but a detailed analysis showed that the best quantitative results were obtained with the 0.5 mCi dose. This method is currently being applied to respiratory gated nave marmosets and NZW rabbits chronically infected with their respective Mtb strains using 0.2mCi/kg, 0.5mCi/kg, 1.0mCi/kg and 2.0mCi/kg. Our analysis of PET images in an immune inhibitor study was successful in detecting disease-related FDG uptake (SUV > 2.5) in abnormal regions in lungs and lymph nodes of rhesus macaques. However, in these animals, not all of the diseased tissue had an elevated FDG uptake. Therefore, we applied an automated method that segregates low- and high-density ranges using a whole lung technique. This approach allowed us to closely and accurately monitor disease changes even if the individual lesions were very difficult to separate in the CT images. We have also collected CTs on several rabbits and rhesus macaques on both the old and new scanners, these images have been quantitatively compared to better define the HU range of abnormal lung densities and in particular TB lesions, this works was featured in the thesis of one the TBIP employees and is being prepared for publication. The use of the new LFER with the marmoset model has been delayed because of the inability of the scanner to collect a lung image within the time we could hold the breath of the animal. Recently we have incorporated new gating hardware to work with new software programing developed for the scanner. This system is expected to allow us to create an artifact-free CT data-set during a selected phase of the animals breathing cycle. Preliminary tests suggest gating has helped improve the animals stability and comfort by alleviating air in the stomach and eliminating the potential for respiratory acidosis from a prolonged mechanical breath hold. To assure the quality of the data we are collecting and to conduct high quality and consistent disease quantification with our imaging modalities, we have established a comprehensive quality control system. To ensure continued consistent performance of the CT subsystem we have implemented the following testing: air calibrations daily; integrated calibration (Offset, Gain, Pixel, Blank, and Blank QC calibration), HU calibration, and air calibration QC monthly; linearity (CATPHAN) and homogeneity yearly, as well as PET/CT registration, spatial resolution, and MTF resolution as needed. To ensure consistent performance of the PET subsystem we have implemented: detector QC (event rate, energy spectrum, time spectrum and crystal matrix QC) daily; accuracy and image quality monthly; uniformity, homogeneity, spatial resolution and sensitivity/efficiency yearly; as well as partial volume, co-registration, and counting performance as needed. To ensure consistent performance of the Dose Calibrator for accurate reproducible dosing of our system is as follows: constancy and accuracy daily; linearity, yearly; as well as geometry calibration with any loss of power or movement as needed. As we work with multiple groups and research models, it is important to keep the experimental data well documented and organized. For each group, we created two network drives to store these data. The purpose of the first drive is to store all the testing data, raw PET/CT scans, a back-up of the reconstructed PET/CT scans, and any preliminary image analysis. The second drive for each group is for storing all data that the PI needs for analysis. This includes the final version of the PET/CT scans, analysis results, in-life data such as bloodwork and anesthesia logs, necropsy documents and images, and histology. This setup allows us to have all the important data in one spot for everyone to reference and a separate limited access-location for the modification-sensitive original image data.
