Comparison of data fusion strategies for automated prostate lesion detection using mpMRI correlated with whole mount histology.

BACKGROUND: In this work, we compare input level, feature level and decision level data fusion techniques for automatic detection of clinically significant prostate lesions (csPCa). METHODS: Multiple deep learning CNN architectures were developed using the Unet as the baseline. The CNNs use both multiparametric MRI images (T2W, ADC, and High b-value) and quantitative clinical data (prostate specific antigen (PSA), PSA density (PSAD), prostate gland volume & gross tumor volume (GTV)), and only mp-MRI images (n = 118), as input. In addition, co-registered ground truth data from whole mount histopathology images (n = 22) were used as a test set for evaluation. RESULTS: The CNNs achieved for early/intermediate / late level fusion a precision of 0.41/0.51/0.61, recall value of 0.18/0.22/0.25, an average precision of 0.13 / 0.19 / 0.27, and F scores of 0.55/0.67/ 0.76. Dice Sorensen Coefficient (DSC) was used to evaluate the influence of combining mpMRI with parametric clinical data for the detection of csPCa. We compared the DSC between the predictions of CNN's trained with mpMRI and parametric clinical and the CNN's trained with only mpMRI images as input with the ground truth. We obtained a DSC of data 0.30/0.34/0.36 and 0.26/0.33/0.34 respectively. Additionally, we evaluated the influence of each mpMRI input channel for the task of csPCa detection and obtained a DSC of 0.14 / 0.25 / 0.28. CONCLUSION: The results show that the decision level fusion network performs better for the task of prostate lesion detection. Combining mpMRI data with quantitative clinical data does not show significant differences between these networks (p = 0.26/0.62/0.85). The results show that CNNs trained with all mpMRI data outperform CNNs with less input channels which is consistent with current clinical protocols where the same input is used for PI-RADS lesion scoring. TRIAL REGISTRATION: The trial was registered retrospectively at the German Register for Clinical Studies (DRKS) under proposal number Nr. 476/14 & 476/19.

Combination of high resolution MRI with 3D-printed needle guides for ex vivo myocardial biopsies.

Magnetic resonance imaging (MRI) provides a multitude of techniques to detect and characterize myocardial infarction. To correlate MRI findings with histology, in most cases terminal animal studies are performed; however, precise extraction and spatial correlation of myocardial tissue samples to MRI image data is difficult. In this proof of concept study, we present a 3D-printing technique to facilitate the extraction of tissue samples from myocardial regions. Initially, seven pig hearts embedded in formaldehyde were imaged on a clinical 3 T system to define biopsy targets on high resolution ex vivo images. Magnitude images and R2*-maps acquired with a 3D multi-echo gradient echo sequence and 0.58 mm isotropic resolution were used to create digital models of the cardiac anatomy. Biopsy guides were 3D-printed to steer the extraction of myocardial samples. In total, 61 tissue samples were extracted with an average offset of the tissue sample location from the target location of 0.59 ± 0.36 mm. This offset was not dependent on the distance of the target point to the epicardial surface. Myocardial tissue could be extracted from all samples. The presented method enables extraction of myocardial tissue samples that are selected by ex vivo MRI with submillimeter precision.

Quality requirements for MRI simulation in cranial stereotactic radiotherapy: a guideline from the German Taskforce "Imaging in Stereotactic Radiotherapy".

Accurate Magnetic Resonance Imaging (MRI) simulation is fundamental for high-precision stereotactic radiosurgery and fractionated stereotactic radiotherapy, collectively referred to as stereotactic radiotherapy (SRT), to deliver doses of high biological effectiveness to well-defined cranial targets. Multiple MRI hardware related factors as well as scanner configuration and sequence protocol parameters can affect the imaging accuracy and need to be optimized for the special purpose of radiotherapy treatment planning. MRI simulation for SRT is possible for different organizational environments including patient referral for imaging as well as dedicated MRI simulation in the radiotherapy department but require radiotherapy-optimized MRI protocols and defined quality standards to ensure geometrically accurate images that form an impeccable foundation for treatment planning. For this guideline, an interdisciplinary panel including experts from the working group for radiosurgery and stereotactic radiotherapy of the German Society for Radiation Oncology (DEGRO), the working group for physics and technology in stereotactic radiotherapy of the German Society for Medical Physics (DGMP), the German Society of Neurosurgery (DGNC), the German Society of Neuroradiology (DGNR) and the German Chapter of the International Society for Magnetic Resonance in Medicine (DS-ISMRM) have defined minimum MRI quality requirements as well as advanced MRI simulation options for cranial SRT.

Simultaneous T2 -weighted real-time MRI of two orthogonal slices.

PURPOSE: MR guidance is used during therapy to detect and compensate for lesion motion. T2 -weighted MRI often has a superior lesion contrast in comparison to T1 -weighted real-time imaging. The purpose of this work was to design a fast T2 -weighted sequence capable of simultaneously acquiring two orthogonal slices, enabling real-time tracking of lesions. METHODS: To generate a T2 contrast in two orthogonal slices simultaneously, a sequence (Ortho-SFFP-Echo) was designed that samples the T2 -weighted spin echo (S- ) signal in a TR-interleaved acquisition of two slices. Slice selection and phase-encoding directions are swapped between the slices, leading to a unique set of spin-echo signal conditions. To minimize motion-related signal dephasing, additional flow-compensation strategies are implemented. In both the abdominal breathing phantom and in vivo experiments, a time series was acquired using Ortho-SSFP-Echo. The centroid of the target was tracked in postprocessing steps. RESULTS: In the phantom, the lesion could be identified and delineated in the dynamic images. In the volunteer experiments, the kidney was visualized with a T2 contrast at a temporal resolution of 0.45 s under free-breathing conditions. A respiratory belt demonstrated a strong correlation with the time course of the kidney centroid in the head-foot direction. A hypointense saturation band at the slice overlap did not inhibit lesion tracking in the semi-automatic postprocessing steps. CONCLUSION: The Ortho-SFFP-Echo sequence delivers real-time images with a T2 -weighted contrast in two orthogonal slices. The sequence allows for simultaneous acquisition, which could be beneficial for real-time motion tracking in radiotherapy or interventional MRI.

In Vivo Metabolic Imaging of [1-13 C]Pyruvate-d3 Hyperpolarized By Reversible Exchange With Parahydrogen.

Metabolic magnetic resonance imaging (MRI) using hyperpolarized (HP) pyruvate is becoming a non-invasive technique for diagnosing, staging, and monitoring response to treatment in cancer and other diseases. The clinically established method for producing HP pyruvate, dissolution dynamic nuclear polarization, however, is rather complex and slow. Signal Amplification By Reversible Exchange (SABRE) is an ultra-fast and low-cost method based on fast chemical exchange. Here, for the first time, we demonstrate not only in vivo utility, but also metabolic MRI with SABRE. We present a novel routine to produce aqueous HP [1-13 C]pyruvate-d3 for injection in 6 minutes. The injected solution was sterile, non-toxic, pH neutral and contained ≈30 mM [1-13 C]pyruvate-d3 polarized to ≈11 % (residual 250 mM methanol and 20 µM catalyst). It was obtained by rapid solvent evaporation and metal filtering, which we detail in this manuscript. This achievement makes HP pyruvate MRI available to a wide biomedical community for fast metabolic imaging of living organisms.

Over 20% Carbon-13 Polarization of Perdeuterated Pyruvate Using Reversible Exchange with Parahydrogen and Spin-Lock Induced Crossing at 50 µT.

Carbon-13 hyperpolarized pyruvate is about to become the next-generation contrast agent for molecular magnetic resonance imaging of cancer and other diseases. Here, efficient and rapid pyruvate hyperpolarization is achieved via signal amplification by reversible exchange (SABRE) with parahydrogen through synergistic use of substrate deuteration, alternating, and static microtesla magnetic fields. Up to 22 and 6% long-lasting 13C polarization (T1 = 3.7 ± 0.25 and 1.7 ± 0.1 min) is demonstrated for the C1 and C2 nuclear sites, respectively. The remarkable polarization levels become possible as a result of favorable relaxation dynamics at the microtesla fields. The ultralong polarization lifetimes will be conducive to yielding high polarization after purification, quality assurance, and injection of the hyperpolarized molecular imaging probes. These results pave the way to future in vivo translation of carbon-13 hyperpolarized molecular imaging probes prepared by this approach.

RF-induced heating of interventional devices at 23.66 MHz.

OBJECTIVE: Low-field MRI systems are expected to cause less RF heating in conventional interventional devices due to lower Larmor frequency. We systematically evaluate RF-induced heating of commonly used intravascular devices at the Larmor frequency of a 0.55 T system (23.66 MHz) with a focus on the effect of patient size, target organ, and device position on maximum temperature rise. MATERIALS AND METHODS: To assess RF-induced heating, high-resolution measurements of the electric field, temperature, and transfer function were combined. Realistic device trajectories were derived from vascular models to evaluate the variation of the temperature increase as a function of the device trajectory. At a low-field RF test bench, the effects of patient size and positioning, target organ (liver and heart) and body coil type were measured for six commonly used interventional devices (two guidewires, two catheters, an applicator and a biopsy needle). RESULTS: Electric field mapping shows that the hotspots are not necessarily localized at the device tip. Of all procedures, the liver catheterizations showed the lowest heating, and a modification of the transmit body coil could further reduce the temperature increase. For common commercial needles no significant heating was measured at the needle tip. Comparable local SAR values were found in the temperature measurements and the TF-based calculations. CONCLUSION: At low fields, interventions with shorter insertion lengths such as hepatic catheterizations result in less RF-induced heating than coronary interventions. The maximum temperature increase depends on body coil design.

Assessing critical temperature dose areas in the kidney by magnetic resonance imaging thermometry in an ex vivo Holmium:YAG laser lithotripsy model.

PURPOSE: We aimed to assess critical temperature areas in the kidney parenchyma using magnetic resonance thermometry (MRT) in an ex vivo Holmium:YAG laser lithotripsy model. METHODS: Thermal effects of Ho:YAG laser irradiation of 14 W and 30 W were investigated in the calyx and renal pelvis of an ex vivo kidney with different laser application times (tL) followed by a delay time (tD) of tL/tD = 5/5 s, 5/10 s, 10/5 s, 10/10 s, and 20/0 s, with irrigation rates of 10, 30, 50, 70, and 100 ml/min. Using MRT, the size of the area was determined in which the thermal dose as measured by the Cumulative Equivalent Minutes (CEM43) method exceeded a value of 120 min. RESULTS: In the calyx, CEM43 never exceeded 120 min for flow rates ≥ 70 ml/min at 14 W, and longer tL (10 s vs. 5 s) lead to exponentially lower thermal affection of tissue (3.6 vs. 21.9 mm2). Similarly at 30 W and ≥ 70 ml/min CEM43 was below 120 min. Interestingly, at irrigation rates of 10 ml/min, tL = 10 s and tD = 10 s CEM43 were observed > 120 min in an area of 84.4 mm2 and 49.1 mm2 at tD = 5 s. Here, tL = 5 s revealed relevant thermal affection of 29.1 mm2 at 10 ml/min. CONCLUSION: We demonstrate that critical temperature dose areas in the kidney parenchyma were associated with high laser power and application times, a low irrigation rate, and anatomical volume of the targeted calyx.

Radio-frequency induced heating of intra-cranial EEG electrodes: The more the colder?

Many neurological disorders are analyzed and treated with implantable electrodes. Many patients with such electrodes have to undergo MRI examinations - often unrelated to their implant - at the risk of radio-frequency induced heating. The number of electrode contact sites of these implants keeps increasing due to improvements in manufacturing and computational algorithms. Electrode grids with multiple receive channels couple to the RF fields present in MRI, but, due to their proximity, a combination of leads has a coupling response which is not a superposition of the individual leads' response. To investigate the problem of RF-induced heating of coupled multi-lead implants, temperature mapping was performed on a set of intra-cranial electroencephalogram (icEEG) electrode grid prototypes with increasing number of contact sites (1-16). Additionally, electric field measurements were used to investigate the radio-frequency heating characteristics of the implants in different media combinations, simulating the device being partially immersed inside the patient. MR measurements show RF-induced heating up to 19.6 K for the single electrode, reducing monotonically with larger number of contact sites to a minimum of 0.9 K for the largest grid. The SAR calculated from temperature measurements agrees well with electric field mapping: The same trend is visible for different insertion lengths, however, the energy dissipated by the whole implant varies with the grid size and insertion length. Thus, in the tested circumstances, a larger electrode number either reduced or had a similar risk of RF induced heating, indicating, that the size of electrode grids is a design parameter, which can be used to change an implants RF response and in turn to reduce the risk of RF induced heating and improve the safety of patient with neuro-implants undergoing MRI examinations.

Counter-historical study of alternative dispersant use in the Deepwater Horizon oil spill response.

Recent completion of oil fate modeling and a mass budget of the Deepwater Horizon (DWH) oil spill allows for a counter-historical study using quantitative Comparative Risk Assessment (CRA) methodology. Novel application of subsea dispersant injection (SSDI) during the response reduced surfacing oil, volatile organic carbon emissions, and oil on shorelines. The effectiveness of that application, and potential alternatives had dispersant not been used or been used more aggressively, were evaluated by modifying and comparing the validated oil fate model under different SSDI strategies. A comparison of mass balance results, exposure metrics, and CRA scoring for Valued Ecological Components (VECs) shows the value of SSDI in achieving risk reduction and tradeoffs that were made. Actual SSDI applied during the DWH oil spill reduced exposures to varying degrees for different VECs. Exposures and relative risks across the ecosystem would have been substantially reduced with more effective SSDI.

Explainable AI for CNN-based prostate tumor segmentation in multi-parametric MRI correlated to whole mount histopathology.

Automatic prostate tumor segmentation is often unable to identify the lesion even if multi-parametric MRI data is used as input, and the segmentation output is difficult to verify due to the lack of clinically established ground truth images. In this work we use an explainable deep learning model to interpret the predictions of a convolutional neural network (CNN) for prostate tumor segmentation. The CNN uses a U-Net architecture which was trained on multi-parametric MRI data from 122 patients to automatically segment the prostate gland and prostate tumor lesions. In addition, co-registered ground truth data from whole mount histopathology images were available in 15 patients that were used as a test set during CNN testing. To be able to interpret the segmentation results of the CNN, heat maps were generated using the Gradient Weighted Class Activation Map (Grad-CAM) method. The CNN achieved a mean Dice Sorensen Coefficient 0.62 and 0.31 for the prostate gland and the tumor lesions -with the radiologist drawn ground truth and 0.32 with whole-mount histology ground truth for tumor lesions. Dice Sorensen Coefficient between CNN predictions and manual segmentations from MRI and histology data were not significantly different. In the prostate the Grad-CAM heat maps could differentiate between tumor and healthy prostate tissue, which indicates that the image information in the tumor was essential for the CNN segmentation.

Improvement of diffusion weighted MRI by practical B0 homogenization for head & neck cancer patients undergoing radiation therapy.

BACKGROUND: MRI is a frequently used tool in radiation therapy planning. For MR-based tumor segmentation, diffusion weighted imaging plays a major role, which can fail due to excessive image artifacts for head and neck cancer imaging. Here, an easy-to-use setup is presented for imaging of head and neck cancer patients in a radiotherapy thermoplastic fixation mask. METHODS: In a prospective head and neck cancer study, MRI data of 29 patients has been acquired at 3 different time points during radiation treatment. The data was analyzed with respect to Nyquist ghosting artifacts in the diffusion images in conventional single shot and readout segmented EPI sequences. For 9 patients, an improved setup with water bags for B0 homogenization was used, and the impact on artifact frequency was analyzed. Additionally, volunteer measurements with B0 fieldmaps are presented. RESULTS: The placement of water bags to the sides of the head during MRI measurements significantly reduces artefacts in diffusion MRI. The number of artifact-free images in readout segmented EPI increased from 74% to 95% of the cases. Volunteer measurements showed a significant increase in B0 homogeneity across slices (head foot direction) as well as within each slice. CONCLUSIONS: The placement of water bags for B0 homogenization is easy to implement, cost-efficient and does not impact patient comfort. Therefore, if very sophisticated soft- or hardware solutions are not present at a given site, or cannot be implemented due to restrictions from the thermoplastic mask, this is an excellent alternative to reduce artifacts in diffusion weighted imaging.

New developments in MRI: System characterization, technical advances and radiotherapy applications.

A Special Issue of Physica Medica - European Journal of Medical Physics, focused on some important points of contact between the world of magnetic resonance and that of medical physics, was published during 2021. This Editorial describes and comments on the content of this Focus Issue, which contains articles from leading groups invited by the Guest Editors.

Supportive Hepatocyte Transplantation after Partial Hepatectomy Enhances Liver Regeneration in a Preclinical Pig Model.

BACKGROUND: Hepatocyte transplantation (HTx) is regarded as a potential treatment modality for various liver diseases including acute liver failure. We developed a preclinical pig model to evaluate if HTx could safely support recovery from liver function impairment after partial hepatectomy. METHODS: Pigs underwent partial hepatectomy with reduction of the liver volume by 50% to induce a transient but significant impairment of liver function. Thereafter, 2 protocols for HTx were evaluated and compared to a control group receiving liver resection only (group 1, n = 5). Portal pressure-controlled HTx was performed either immediately after surgery (group 2, n = 6) or 3 days postoperatively (group 3, n = 5). In all cases, liver regeneration was monitored by conventional laboratory tests and the novel noninvasive maximum liver function capacity (LiMAx) test with a follow-up of 4 weeks. RESULTS: Partial hepatectomy significantly impaired liver function according to conventional liver function tests as well as LiMAx in all groups. A mean of 4.10 ± 1.1 × 108 and 3.82 ± 0.7 × 108 hepatocytes were transplanted in groups 2 and 3, respectively. All animals remained stable with respect to vital parameters during and after HTx. The animals in group 2 showed enhanced liver regeneration as observed by mean postoperative LiMAx values (621.5 vs. 331.3 µg/kg/h on postoperative day 7; p < 0.001) whereas HTx in group 3 led to a significant increase in mean liver-specific coagulation factor VII (112.2 vs. 54.0% on postoperative day 7; p = 0.003) compared to controls (group 1), respectively. In both experimental groups, thrombotic material was observed in the portal veins and pulmonary arteries on histology, despite the absence of clinical symptoms. CONCLUSION: HTx can be performed safely and effectively immediately after a partial (50%) hepatectomy as well as 3 days postoperatively, with comparable results regarding the enhancement of liver function and regeneration.

Nafcillin-Induced Hepatic Injury: A Case Report and Literature Review.

BACKGROUND: Drug-induced liver injury (DILI) is the most common cause of acute liver failure in the Western world. While it requires a diagnosis of exclusion, it is exceedingly prevalent in patients taking multiple hepatotoxic agents, the foremost of which are antibiotics, followed by herbal and dietary supplements. Below we will discuss a case of nafcillin-induced liver injury suggested by a thorough work-up and rule-out of other hepatic and biliary pathologies. CASE PRESENTATION:  We report the case of a 66-year-old white male who presented with painless jaundice. Clinical, laboratory and radiographic features demonstrated a cholestatic pattern of liver injury without significant abnormalities in the biliary tract. All workup for viral hepatitis and autoimmune diseases with liver involvement was negative. Liver biopsy showed acute necro-inflammatory changes suggestive of drug-induced liver injury. The patient had received 18 days of IV nafcillin for blood culture positive methicillin-susceptible Staphylococcus aureus (MSSA) four weeks prior to his presentation. He showed clinical and laboratory improvement of his liver functions with supportive care only. CONCLUSION: Nafcillin is a safe and effective antibiotic for the treatment of methicillin-susceptible Staphylococcal infections. However, physicians and prescribing healthcare professionals should be aware of the rare, but serious side effects, especially one of drug-induced liver injury with emphasis on the need for early cessation of nafcillin if liver function abnormalities develop.

Passive needle guide tracking with radial acquisition and phase-only cross-correlation.

PURPOSE: Acceleration of a passive tracking sequence based on phase-only cross-correlation (POCC) using radial undersampling. METHODS: The phase-only cross-correlation (POCC) algorithm allows passive tracking of interventional instruments in real-time. In a POCC sequence, two cross-sectional images of a needle guide with a positive MR contrast are continuously acquired from which the instrument trajectory is calculated. Conventional Cartesian imaging for tracking is very time consuming; here, a higher temporal resolution is achieved using a highly undersampled radial acquisition together with a modified POCC algorithm that incorporates the point-spread-function. Targeting and needle insertion is performed in two phantom experiments with 16 fiducial targets, each using 4 and 16 radial projections for passive tracking. Additionally, targeting of eight deep lying basivertebral veins in the lumbar spines is performed for in vivo proof-of-application with four radial projections for needle guide tracking. RESULTS: The radially undersampled POCC sequence yielded in the phantom experiments a lateral targeting accuracy of 1.1 ± 0.4 mm and 1.0 ± 0.5 mm for 16 and 4 radial projections, respectively, without any statistically significant difference. In the in vivo application, a mean targeting duration of 62 ± 13 s was measured. CONCLUSION: Radial undersampling can drastically reduce the acquisition time for passive tracking in a POCC sequences for MR-guided needle interventions without compromising the targeting accuracy.

Isotropic Expansion of the Intraprostatic Gross Tumor Volume of Primary Prostate Cancer Patients Defined in MRI-A Correlation Study With Whole Mount Histopathological Information as Reference.

INTRODUCTION: An accurate delineation of the intraprostatic gross tumor volume (GTV) is of importance for focal treatment in patients with primary prostate cancer (PCa). Multiparametric MRI (mpMRI) is the standard of care for lesion detection but has been shown to underestimate GTV. This study investigated how far the GTV has to be expanded in MRI in order to reach concordance with the histopathological reference and whether this strategy is practicable in clinical routine. PATIENTS AND METHODS: Twenty-two patients with planned prostatectomy and preceded 3 Tesla mpMRI were prospectively examined. After surgery, PCa contours delineated on histopathological slides (GTV-Histo) were superimposed on MRI using ex-vivo imaging as support for co-registration. According to the PI-RADSv2 classification, GTV was manually delineated in MRI (GTV-MRI) by two experts in consensus. For volumetric analysis, we compared GTV-MRI and GTV-Histo. Subsequently, we isotropically enlarged GTV-MRI in 1 mm increments within the prostate and also compared those with GTV-Histo regarding the absolute volumes. For evaluating the spatial accuracy, we considered the coverage ratio of GTV-Histo, the Sørensen-Dice coefficient (DSC), as well as the contact with the urethra. RESULTS: In 19 of 22 patients MRI underestimated the intraprostatic tumor volume compared to histopathological reference: median GTV-Histo (4.7 cm3, IQR: 2.5-18.8) was significantly (p<0.001) lager than median GTV-MRI (2.6 cm3, IQR: 1.2-6.9). A median expansion of 1 mm (range: 0-4 mm) adjusted the initial GTV-MRI to at least the volume of GTV-Histo (GTVexp-MRI). Original GTV-MRI and expansion with 1, 2, 3, and 4 mm covered in median 39% (IQR: 2%-78%), 62% (10%-91%), 70% (15%-95%), 80% (21-100), 87% (25%-100%) of GTV-Histo, respectively. Best DSC (median: 0.54) between GTV-Histo and GTV-MRI was achieved by median expansion of 2 mm. The urethra was covered by initial GTVs-MRI in eight patients (36%). After applying an expansion with 2 mm the urethra was covered in one more patient by GTV-MRI. CONCLUSION: Using histopathology as reference, we demonstrated that MRI underestimates intraprostatic tumor volume. A 2 mm-expansion may improve accurate GTV-delineation while respecting the balance between histological tumor coverage and overtreatment.

Predicting Biochemical Failure in Irradiated Patients With Prostate Cancer by Tumour Volume Measured by Multiparametric MRI.

BACKGROUND/AIM: We examined the prognostic value of intraprostatic gross tumour volume (GTV) as measured by multiparametric MRI (mpMRI) in patients with prostate cancer following (primary) external beam radiation therapy (EBRT). PATIENTS AND METHODS: In a retrospective monocentric study, we analysed patients with prostate cancer (PCa) after EBRT. GTV was delineated in pre-treatment mpMRI (GTV-MRI) using T2-weighted images. Cox-regression analyses were performed considering biochemical failure recurrence-free survival (BRFS) as outcome variable. RESULTS: Among 131 patients, after a median follow-up of 57 months, biochemical failure occurred in 27 (21%). GTV-MRI was not correlated with % of positive biopsy cores, Gleason score and initial PSA (all r<0.2) and only moderately correlated with cT stage (r=0.32). In univariate analysis, cT stage, Gleason score and GTV-MRI were higher in subjects with shorter BRFS (p<0.05). GTV-MRI remained a significant predictor for BRFS in multivariate analyses, independent of Gleason score and cT stage. CONCLUSION: GTV, defined using mpMRI, provides incremental prognostic value for BRFS, independent of established risk factors. This supports the implementation of imaging-based GTV for risk-stratification, although further validation is needed.

Joint Imaging Platform for Federated Clinical Data Analytics.

PURPOSE: Image analysis is one of the most promising applications of artificial intelligence (AI) in health care, potentially improving prediction, diagnosis, and treatment of diseases. Although scientific advances in this area critically depend on the accessibility of large-volume and high-quality data, sharing data between institutions faces various ethical and legal constraints as well as organizational and technical obstacles. METHODS: The Joint Imaging Platform (JIP) of the German Cancer Consortium (DKTK) addresses these issues by providing federated data analysis technology in a secure and compliant way. Using the JIP, medical image data remain in the originator institutions, but analysis and AI algorithms are shared and jointly used. Common standards and interfaces to local systems ensure permanent data sovereignty of participating institutions. RESULTS: The JIP is established in the radiology and nuclear medicine departments of 10 university hospitals in Germany (DKTK partner sites). In multiple complementary use cases, we show that the platform fulfills all relevant requirements to serve as a foundation for multicenter medical imaging trials and research on large cohorts, including the harmonization and integration of data, interactive analysis, automatic analysis, federated machine learning, and extensibility and maintenance processes, which are elementary for the sustainability of such a platform. CONCLUSION: The results demonstrate the feasibility of using the JIP as a federated data analytics platform in heterogeneous clinical information technology and software landscapes, solving an important bottleneck for the application of AI to large-scale clinical imaging data.

Convolutional neural networks for head and neck tumor segmentation on 7-channel multiparametric MRI: a leave-one-out analysis.

BACKGROUND: Automatic tumor segmentation based on Convolutional Neural Networks (CNNs) has shown to be a valuable tool in treatment planning and clinical decision making. We investigate the influence of 7 MRI input channels of a CNN with respect to the segmentation performance of head&neck cancer. METHODS: Head&neck cancer patients underwent multi-parametric MRI including T2w, pre- and post-contrast T1w, T2*, perfusion (ktrans, ve) and diffusion (ADC) measurements at 3 time points before and during radiochemotherapy. The 7 different MRI contrasts (input channels) and manually defined gross tumor volumes (primary tumor and lymph node metastases) were used to train CNNs for lesion segmentation. A reference CNN with all input channels was compared to individually trained CNNs where one of the input channels was left out to identify which MRI contrast contributes the most to the tumor segmentation task. A statistical analysis was employed to account for random fluctuations in the segmentation performance. RESULTS: The CNN segmentation performance scored up to a Dice similarity coefficient (DSC) of 0.65. The network trained without T2* data generally yielded the worst results, with ΔDSCGTV-T = 5.7% for primary tumor and ΔDSCGTV-Ln = 5.8% for lymph node metastases compared to the network containing all input channels. Overall, the ADC input channel showed the least impact on segmentation performance, with ΔDSCGTV-T = 2.4% for primary tumor and ΔDSCGTV-Ln = 2.2% respectively. CONCLUSIONS: We developed a method to reduce overall scan times in MRI protocols by prioritizing those sequences that add most unique information for the task of automatic tumor segmentation. The optimized CNNs could be used to aid in the definition of the GTVs in radiotherapy planning, and the faster imaging protocols will reduce patient scan times which can increase patient compliance. TRIAL REGISTRATION: The trial was registered retrospectively at the German Register for Clinical Studies (DRKS) under register number DRKS00003830 on August 20th, 2015.