Adam Kesner

On transcending the impasse of respiratory motion correction applications in routine clinical imaging - a consideration of a fully automated data driven motion control framework

Positron emission tomography (PET) is increasingly used for the detection, characterization, and follow-up of tumors located in the thorax. However, patient respiratory motion presents a unique limitation that hinders the application of high-resolution PET technology for this type of imaging. Efforts to transcend this limitation have been underway for more than a decade, yet PET remains for practical considerations a modality vulnerable to motion-induced image degradation. Respiratory motion control is not employed in routine clinical operations. In this article, we take an opportunity to highlight some of the recent advancements in data-driven motion control strategies and how they may form an underpinning for what we are presenting as a fully automated data-driven motion control framework. This framework represents an alternative direction for future endeavors in motion control and can conceptually connect individual focused studies with a strategy for addressing big picture challenges and goals.

Respiratory Gated PET Derived in a Fully Automated Manner From Raw PET Data

Respiratory motion in PET degrades image quality and limits detectability of small or low-contrast lesions. Although image quality can be improved using respiratory-gating, this adds to the complexity and expense of acquiring PET data. We aimed to develop a data-driven method, based on individual voxel signal fluctuations, for accomplishing electronic respiratory gating of clinical PET data, requiring no additional hardware or end-user input. We tested our methods using both simulated PET scans and actual human PET acquisitions. For the simulations, our methods correctly identified the start frame of each respiratory cycle defined for the phantom. Resultant gated images demonstrated improved effective resolution and increased measured uptake for lesions located in the thorax. For human PET data, we were able to recover respiratory phase information with a high signal-to-noise ratio. We report here a method to achieve fully automated voxel-based respiratory gating of PET images, without the need for gating hardware or additional user input, capable of improving effective resolution and increasing lesion detectability.

Validation of Software Gating: A Practical Technology for Respiratory Motion Correction in PET

Purpose To assess the performance of hardware- and software-gating technologies in terms of qualitative and quantitative characteristics of respiratory motion in positron emission tomography (PET) imaging. Materials and Methods Between 2010 and 2013, 219 fluorine 18 fluorodeoxyglucose PET examinations were performed in 116 patients for assessment of pulmonary nodules. All patients provided informed consent in this institutional review board-approved study. Acquisitions were reconstructed as respiratory-gated images by using hardware-derived respiratory triggers and software-derived signal (via an automated postprocessing method). Asymmetry was evaluated in the joint distribution of reader preference, and linear mixed models were used to evaluate differences in outcomes according to gating type. Results In blind reviews of reconstructed gated images, software was selected as superior 16.9% of the time (111 of 657 image sets; 95% confidence interval [CI]: 14.0%, 19.8%), and hardware was selected as superior 6.2% of the time (41 of 657 image sets; 95% CI: 4.4%, 8.1%). Of the image sets, 76.9% (505 of 657; 95% CI: 73.6%, 80.1%) were judged as having indistinguishable motion quality. Quantitative analysis demonstrated that the two gating strategies exhibited similar performance, and the performance of both was significantly different from that of nongated images. The mean increase ± standard deviation in lesion maximum standardized uptake value was 42.2% ± 38.9 between nongated and software-gated images, and lesion full width at half maximum values decreased by 9.9% ± 9.6. Conclusion Compared with vendor-supplied respiratory-gating hardware methods, software gating performed favorably, both qualitatively and quantitatively. Fully automated gating is a feasible approach to motion correction of PET images. (©) RSNA, 2016 Online supplemental material is available for this article.

Small-Animal PET/CT for Monitoring the Development and Response to Chemotherapy of Thymic Lymphoma in Trp53−/− Mice

Transgenic mouse models of human cancers represent one of the most promising approaches to elucidate clinically relevant mechanisms of action and provide insights into the treatment efficacy of new antitumor drugs. The use of Trp53 transgenic mice (Trp53 knockout [Trp53−/−] mice) for these kinds of studies is, so far, restricted by limitations in detecting developing tumors and the lack of noninvasive tools for monitoring tumor growth, progression, and treatment response. Methods: We hypothesized that quantitative small-animal PET with 18F-FDG was able to detect the onset and location of tumor development, follow tumor progression, and monitor response to chemotherapy. To test these hypotheses, C57BL/6J Trp53−/− mice underwent longitudinal small-animal PET during lymphoma development and gemcitabine treatment. Trp53 wild-type (Trp53+/+) mice were used as controls, and histology after full necropsy served as the gold standard. Results: In Trp53+/+ mice, the thymic standardized uptake value (SUV) did not exceed 1.0 g/mL, with decreasing 18F-FDG uptake over time. Conversely, all Trp53−/− mice that developed thymic lymphoma showed increasing thymic glucose metabolism, with a mean SUV doubling time of 9.0 wk (range, 6.0–17.5 wk). Using an SUV of 3.0 g/mL as a criterion provided a sensitivity of 78% and a specificity of 100% for the detection of thymic lymphoma. Treatment monitoring with 18F-FDG PET correctly identified all histologic responses and relapses to gemcitabine. Conclusion: 18F-FDG small-animal PET can be used to visualize onset and progression of thymic lymphomas in Trp53−/− mice and monitor response to chemotherapy. Thus, 18F-FDG small-animal PET provides an in vivo means to assess intervention studies in the Trp53 transgenic mouse model.

Respiratory gated PET derived from raw PET data

Respiratory motion in PET degrades images and limits detectability of small or low- contrast lesions. Although image quality can be improved using respiratory-gating hardware, this adds to the complexity and expense of acquiring PET data. We aimed to develop a data-driven method, based on individual voxel signal fluctuations, for accomplishing electronic respiratory gating of PET data acquired in a clinically practical manner, requiring no additional hardware or end-user input. We tested our methods using both simulated PET scans, as well as actual human PET acquisitions. For the simulations, our methods correctly identified the start frame of each respiratory cycle defined for the phantom. Resultant gated images demonstrated improved effective resolution and increased SUV uptake for lesions scattered in the thorax. For the human PET data, we were able to recover respiratory phase information with a large signal-to-noise ratio. Fully automated voxel-based respiratory gating of PET images may be achieved without the need for gating hardware or additional user input, in a manner capable of improving effective resolution and increasing lesion detectability.

Evaluation of Image Noise in Respiratory Gated PET

The aim of this study was to quantify image noise and signal recovery in respiratory gated PET. A Jaszczak phantom filled with 18F was placed on a custom built motion platform. Different source to background activity ratios were used. An Anzai belt, a surface tension monitoring device, was strapped around the phantom to track the motion and to trigger the gated PET cycle. Data were acquired into 12 bins throughout one gating cycle. The binned data were also summed to produce image sets representing acquisitions with different numbers of gates, including a non-gated image set. The image noise was estimated using the bootstrap method. Images were generated from 100 sinogram replicates and reconstructed using ordered subsets-expectation maximization (OSEM), 4 iterations and 8 subsets. From the reconstructed image replicates, mean and standard deviation images were created, from which the average image signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of each sphere were calculated. The SNR and CNR were evaluated as a function of the number of gates. The SNR and CNR result in the expected Poisson limited correlation with the number of gates for the larger lesions. Because of the motion, the CNR calculated from the images produced with no or few gates is nearly a factor of 2 less than the expected value for the 3 smallest spheres. As the number of gates increases, the CNR correlates with the expected values. Beyond 6 gates, image noise dominates over any signal improvement, which is reflected in the low CNR values of the smallest spheres. The results of this study show that gating can provide improvement in signal recovery with minimal loss of CNR for small, moving lesions

Time-course of effects of external beam radiation on [18F]FDG uptake in healthy tissue and bone marrow.

The utility of PET for monitoring responses to radiation therapy have been complicated by metabolically active processes in surrounding normal tissues. We examined the time‐course of [18F]FDG uptake in normal tissues using small animal‐dedicated PET during the 2 month period following external beam radiation. Four mice received 12 Gy of external beam radiation, in a single fraction to the left half of the body. Small animal [18F]FDG‐PET scans were acquired for each mouse at 0 (pre‐radiation), 1, 2, 3, 4, 5, 8, 12, 19, 24, and 38 days following irradiation. [18F]FDG activity in various tissues was compared between irradiated and non‐irradiated body halves before, and at each time point after irradiation. Radiation had a significant impact on [18F]FDG uptake in previously healthy tissues, and time‐course of effects differed in different types of tissues. For example, liver tissue demonstrated increased uptake, particularly over days 3–12, with the mean left to right uptake ratio increasing 52% over mean baseline values (p<0.0001). In contrast, femoral bone marrow uptake demonstrated decreased uptake, particularly over days 2–8, with the mean left to right uptake ratio decreasing 26% below mean baseline values (p=0.0005). Significant effects were also seen in lung and brain tissue. Radiation had diverse effects on [18F]FDG uptake in previously healthy tissues. These kinds of data may help lay groundwork for a systematically acquired database of the time‐course of effects of radiation on healthy tissues, useful for animal models of cancer therapy imminently, as well as interspecies extrapolations pertinent to clinical application eventually. PACs Number: 87.50.‐a

Small Data: a ubiquitous, yet untapped, resource for low cost imaging innovation

PET, conventional nuclear imaging, and most contemporary medical imaging modalities are inherently digital technologies. Over the last several decades, there has been a transformative evolution of the digital computing landscape with respect to speed, cost of storage, infrastructure, and available expertise. However, our use of data, and in fact our whole understanding of the role of data in relation to emission imaging, has remained relatively unchanged. If we take a moment to reflect on this resource, generated ubiquitously in our daily imaging procedures, we can recognize that we have the capacity to support information use beyond the present convention and that the raw data provided by nuclear imaging studies can be tapped to fuel innovation. Our general understanding of image data is that it exists in DICOM-format images, essentially analogous to film and representing a quantity of source signal distributed in space. However, the signals and information used to create these images in nuclear medicine originate in a much denser form; our imaging machines capture highly detailed time, location, and energy information for individual decay events. The current practice in PET, for example, is to truncate this information using assumptions and reconstruction techniques so as to provide a representation of tracer emissions distributed in recognizable Cartesian space. This process of biodistribution–representative image generation has essentially defined nuclear imaging for half a century. The procedure of truncating (unused) information is heavily ingrained in our practice likely because, for most of the field’s existence, it has been expensive and impractical to save raw acquisition data. The costs associated with saving data have never been a static consideration. In 1980, a gigabyte of data cost

Frequency based gating: An alternative, conformal, approach to 4D PET data utilization

600,000 (1) (approximate value, inflation-adjusted), in 1990 that cost went down to

On noting the achievements and future potential of data-driven gating for respiratory motion correction in PET imaging.

15,000, in 2016 it went down to