Personalized Pressure Conditions and Calibration for a Predictive Computational Model of Coronary and Myocardial Blood Flow.

Predictive modeling of hyperemic coronary and myocardial blood flow (MBF) greatly supports diagnosis and prognostic stratification of patients suffering from coronary artery disease (CAD). In this work, we propose a novel strategy, using only readily available clinical data, to build personalized inlet conditions for coronary and MBF models and to achieve an effective calibration for their predictive application to real clinical cases. Experimental data are used to build personalized pressure waveforms at the aortic root, representative of the hyperemic state and adapted to surrogate the systolic contraction, to be used in computational fluid-dynamics analyses. Model calibration to simulate hyperemic flow is performed in a "blinded" way, not requiring any additional exam. Coronary and myocardial flow simulations are performed in eight patients with different clinical conditions to predict FFR and MBF. Realistic pressure waveforms are recovered for all the patients. Consistent pressure distribution, blood velocities in the large arteries, and distribution of MBF in the healthy myocardium are obtained. FFR results show great accuracy with a per-vessel sensitivity and specificity of 100% according to clinical threshold values. Mean MBF shows good agreement with values from stress-CTP, with lower values in patients with diagnosed perfusion defects. The proposed methodology allows us to quantitatively predict FFR and MBF, by the exclusive use of standard measures easily obtainable in a clinical context. This represents a fundamental step to avoid catheter-based exams and stress tests in CAD diagnosis.

Amplifying the Effects of Contrast Agents on Magnetic Resonance Images Using a Deep Learning Method Trained on Synthetic Data.

OBJECTIVES: Artificial intelligence (AI) methods can be applied to enhance contrast in diagnostic images beyond that attainable with the standard doses of contrast agents (CAs) normally used in the clinic, thus potentially increasing diagnostic power and sensitivity. Deep learning-based AI relies on training data sets, which should be sufficiently large and diverse to effectively adjust network parameters, avoid biases, and enable generalization of the outcome. However, large sets of diagnostic images acquired at doses of CA outside the standard-of-care are not commonly available. Here, we propose a method to generate synthetic data sets to train an "AI agent" designed to amplify the effects of CAs in magnetic resonance (MR) images. The method was fine-tuned and validated in a preclinical study in a murine model of brain glioma, and extended to a large, retrospective clinical human data set. MATERIALS AND METHODS: A physical model was applied to simulate different levels of MR contrast from a gadolinium-based CA. The simulated data were used to train a neural network that predicts image contrast at higher doses. A preclinical MR study at multiple CA doses in a rat model of glioma was performed to tune model parameters and to assess fidelity of the virtual contrast images against ground-truth MR and histological data. Two different scanners (3 T and 7 T, respectively) were used to assess the effects of field strength. The approach was then applied to a retrospective clinical study comprising 1990 examinations in patients affected by a variety of brain diseases, including glioma, multiple sclerosis, and metastatic cancer. Images were evaluated in terms of contrast-to-noise ratio and lesion-to-brain ratio, and qualitative scores. RESULTS: In the preclinical study, virtual double-dose images showed high degrees of similarity to experimental double-dose images for both peak signal-to-noise ratio and structural similarity index (29.49 dB and 0.914 dB at 7 T, respectively, and 31.32 dB and 0.942 dB at 3 T) and significant improvement over standard contrast dose (ie, 0.1 mmol Gd/kg) images at both field strengths. In the clinical study, contrast-to-noise ratio and lesion-to-brain ratio increased by an average 155% and 34% in virtual contrast images compared with standard-dose images. Blind scoring of AI-enhanced images by 2 neuroradiologists showed significantly better sensitivity to small brain lesions compared with standard-dose images (4.46/5 vs 3.51/5). CONCLUSIONS: Synthetic data generated by a physical model of contrast enhancement provided effective training for a deep learning model for contrast amplification. Contrast above that attainable at standard doses of gadolinium-based CA can be generated through this approach, with significant advantages in the detection of small low-enhancing brain lesions.

Tumor vascular remodeling by thrombospondin-1 enhances drug delivery and antineoplastic activity.

The disorganized and inefficient tumor vasculature is a major obstacle to the delivery and efficacy of antineoplastic treatments. Antiangiogenic agents can normalize the tumor vessels, improving vessel function and boosting the distribution and activity of chemotherapy. The type III repeats (T3R) domain of thrombospondin-1 contains different potential antiangiogenic sequences. We therefore hypothesized that it might affect the tumor vasculature. Ectopic expression of the T3R domain by the tumor cells or by the host, or administration of recombinant T3R, delayed the in vivo growth of experimental tumors. Tumors presented marked reorganization of the vasculature, with abundant but smaller vessels, associated with substantially less necrosis. Mechanistically, the use of truncated forms of the domain, containing different active sequences, pointed to the FGF2/FGFR/ERK axis as a target for T3R activity. Along with reduced necrosis, the expression of T3R promoted tumor distribution of chemotherapy (paclitaxel), with a higher drug concentration and more homogeneous distribution, as assessed by HPLC and MALDI imaging mass spectrometry. T3R-expressing tumors were more responsive to paclitaxel and cisplatin. This study shows that together with its known role as a canonical inhibitor of angiogenesis, thrombospondin-1 can also remodel tumor blood vessels, affecting the morphological and functional properties of the tumor vasculature. The ability of T3R to reduce tumor growth and improve the response to chemotherapy opens new perspectives for therapeutic strategies based on T3R to be used in combination therapies.

AIforCOVID: Predicting the clinical outcomes in patients with COVID-19 applying AI to chest-X-rays. An Italian multicentre study.

Recent epidemiological data report that worldwide more than 53 million people have been infected by SARS-CoV-2, resulting in 1.3 million deaths. The disease has been spreading very rapidly and few months after the identification of the first infected, shortage of hospital resources quickly became a problem. In this work we investigate whether artificial intelligence working with chest X-ray (CXR) scans and clinical data can be used as a possible tool for the early identification of patients at risk of severe outcome, like intensive care or death. Indeed, further to induce lower radiation dose than computed tomography (CT), CXR is a simpler and faster radiological technique, being also more widespread. In this respect, we present three approaches that use features extracted from CXR images, either handcrafted or automatically learnt by convolutional neuronal networks, which are then integrated with the clinical data. As a further contribution, this work introduces a repository that collects data from 820 patients enrolled in six Italian hospitals in spring 2020 during the first COVID-19 emergency. The dataset includes CXR images, several clinical attributes and clinical outcomes. Exhaustive evaluation shows promising performance both in 10-fold and leave-one-centre-out cross-validation, suggesting that clinical data and images have the potential to provide useful information for the management of patients and hospital resources.

Detection of an Autism EEG Signature From Only Two EEG Channels Through Features Extraction and Advanced Machine Learning Analysis.

BACKGROUND AND OBJECTIVE: In 2 previous studies, we have shown the ability of special machine learning systems applied to standard EEG data in distinguishing children with autism spectrum disorder (ASD) from non-ASD children with an overall accuracy rate of 100% and 98.4%, respectively. Since the equipment routinely available in neonatology units employ few derivations, we were curious to check if just 2 derivations were enough to allow good performance in the same cases of the above-mentioned studies. METHODS: A continuous segment of artifact-free EEG data lasting 1 minute in ASCCI format from C3 and C4 EEG channels present in 2 previous studies, was used for features extraction and subsequent analyses with advanced machine learning systems. A features extraction software package (Python tsfresh) applied to time-series raw data derived 1588 quantitative features. A special hybrid system called TWIST (Training with Input Selection and Testing), coupling an evolutionary algorithm named Gen-D and a backpropagation neural network, was used to subdivide the data set into training and testing sets as well as to select features yielding the maximum amount of information after a first variable selection performed with linear correlation index threshold. RESULTS: After this intelligent preprocessing, 12 features were extracted from C3-C4 time-series of study 1 and 36 C3-C4 time-series of study 2 representing the EEG signature. Acting on these features the overall accuracy predictive capability of the best artificial neural network acting as a classifier in deciphering autistic cases from typicals (study 1) and other neuropsychiatric disorders (study 2) resulted in 100 % for study 1 and 94.95 % for study 2. CONCLUSIONS: The results of this study suggest that also a minor part of EEG contains precious information useful to detect autism if treated with advanced computational algorithms. This could allow in the future to use standard EEG from newborns to check if the ASD signature is already present at birth.

Preliminary safety and imaging efficacy of the near-infrared fluorescent contrast agent DA364 during fluorescence-guided surgery in dogs with spontaneous superficial tumors.

Tumor-targeting contrast agents may facilitate resection of solid neoplasms during fluorescence-guided surgery. Preliminary safety and imaging efficacy of the near-infrared fluorescent probe DA364 were evaluated during surgical resection of spontaneous solid tumors in 24 dogs. Intra-operative imaging was performed in situ and on excised specimens to evaluate fluorescence intensities of tumor and adjacent tissues. After standard-of-care tumor resection, the wound bed was imaged again, and additional tissue was excised if residual fluorescence was detected. DA364 was well tolerated after intravenous administration. The median tumor-to-background ratio in situ for mammary tumors, mast cell tumors and sarcomas was 1.8 (range 1.2-3.9), 2.2 (range 1.0-5.6), and 4.2 (range 2.0-4.3), respectively. Qualitative intra-operative tumor identification was feasible in half of the cases. Remaining fluorescence was detected in four wound beds that contained residual disease, and in11 tumor-free wound beds, confirmed by histopathology. Overall, DA364 did not raise safety concerns and showed accumulation in different types of spontaneous tumors, showing potential to pinpoint residual disease. Larger clinical trials are necessary to select accurate dosing and imaging protocols for specific indications to evaluate the sensitivity and specificity of the agent.

Fluorescence-guided surgery using indocyanine green in dogs with superficial solid tumours.

BACKGROUND: Near-infrared fluorescence (NIRF) imaging is a relatively novel technique that can aid surgeons during intraoperative tumour identification. METHODS: Nine canine oncology patients (five mammary gland tumours, three mast cell tumours and one melanoma) received intravenous indocyanine green (ICG). After 24 hours, tumours were resected and fluorescence intensities of tumours and surroundings were evaluated. Additional wound bed tissue was resected if residual fluorescence was present after tumour resection. Ex vivo, fluorescence-guided dissection was performed to separate tumour from surrounding tissue. RESULTS: Intraoperative NIRF-guided tumour delineation was feasible in four out of nine dogs. Wound bed imaging after tumour removal identified nine additional fluorescent lesions, of which four contained tumour tissue. One of these four true positive in vivo lesions was missed by standard-of-care inspection. Ex vivo fluorescence-guided tumour dissection showed a sensitivity of 72 per cent and a specificity of 80 per cent in discriminating between tumour and surrounding tissue. CONCLUSION: The value of ICG for intraoperative tumour delineation seems more limited than originally thought. Although NIRF imaging using ICG did identify remaining tumour tissue in the wound bed, a high false positive rate was also observed.

A machine learning approach for differentiating malignant from benign enhancing foci on breast MRI.

BACKGROUND: Differentiate malignant from benign enhancing foci on breast magnetic resonance imaging (MRI) through radiomic signature. METHODS: Forty-five enhancing foci in 45 patients were included in this retrospective study, with needle biopsy or imaging follow-up serving as a reference standard. There were 12 malignant and 33 benign lesions. Eight benign lesions confirmed by over 5-year negative follow-up and 15 malignant histopathologically confirmed lesions were added to the dataset to provide reference cases to the machine learning analysis. All MRI examinations were performed with a 1.5-T scanner. One three-dimensional T1-weighted unenhanced sequence was acquired, followed by four dynamic sequences after intravenous injection of 0.1 mmol/kg of gadobenate dimeglumine. Enhancing foci were segmented by an expert breast radiologist, over 200 radiomic features were extracted, and an evolutionary machine learning method ("training with input selection and testing") was applied. For each classifier, sensitivity, specificity and accuracy were calculated as point estimates and 95% confidence intervals (CIs). RESULTS: A k-nearest neighbour classifier based on 35 selected features was identified as the best performing machine learning approach. Considering both the 45 enhancing foci and the 23 additional cases, this classifier showed a sensitivity of 27/27 (100%, 95% CI 87-100%), a specificity of 37/41 (90%, 95% CI 77-97%), and an accuracy of 64/68 (94%, 95% CI 86-98%). CONCLUSION: This preliminary study showed the feasibility of a radiomic approach for the characterisation of enhancing foci on breast MRI.

Novel Gastrin-Releasing Peptide Receptor Targeted Near-Infrared Fluorescence Dye for Image-Guided Surgery of Prostate Cancer.

PURPOSE: Prostate cancer (PCa), the most widespread male cancer in western countries, is generally eradicated by surgery, especially if localized. However, during surgical procedures, it is not always possible to identify malignant tissues by visual inspection. Among the possible consequences, there is the formation of positive surgical margins, often associated with recurrence. In this work, the gastrin-releasing peptide receptor (GRPR), overexpressed in the prostatic carcinoma and not in healthy tissues or in benign hyperplasia (BPH), is proposed as target molecule to design a novel near-infrared fluorescent (NIRF) probe for image-guided prostatectomy. PROCEDURES: The NIRF dye Sulfo-Cy5.5 was conjugated to a Bombesin-like peptide (BBN), targeting GRPR. The final product, called BBN-Cy5.5, was characterized and tested in vitro on PC-3, DU145, and LnCAP cell lines, using unconjugated Sulfo-Cy5.5 as control. In vivo biodistribution studies were performed by optical imaging in PC-3 tumor-bearing and healthy mice. Finally, simulation of the surgical protocol was carried out. RESULTS: BBN-Cy5.5 showed high water solubility and a good relative quantum yield. The ability of the probe to recognize the GRPR, highly expressed in PC-3 cells, was tested both in vitro and in vivo, where a significant tumor accumulation was achieved 24 h post-injection. Furthermore, a distinguishable fluorescent signal was visible in mice bearing PCa, when the surgery was simulated. By contrast, low signal was found in healthy or BPH-affected mice. CONCLUSIONS: This work proposes a new NIRF probe ideal to target GRPR, biomarker of PCa. The promising data obtained suggest that the dye could allow the real-time intraoperative visualization of prostate cancer.

Environment-specific spectral modeling: A new tool for the analysis of biological specimens.

The recent discovery of fluorescent dyes for improving pathologic tissues identification has highlighted the need of robust methods for performance validation especially in the field of fluorescence-guided surgery. Optical imaging of excised tissue samples is the reference tool to validate the association between dyes localization and the underlying histology in a controlled environment. Spectral unmixing may improve the validation process discriminating dye from endogenous signal. Here, an innovative spectral modeling approach that weights the spectral shifts associated with changes in chemical environment is described. The method is robust against spectral shift variations and its application leads to unbiased spectral weights estimates as demonstrated by numerical simulations. Finally, spectral shifts values computed pixel-wise from spectral images are used to display additional information with potential diagnostic value.

Gadolinium presence, MRI hyperintensities, and glucose uptake in the hypoperfused rat brain after repeated administrations of gadodiamide.

PURPOSE: The discussed topic about gadolinium-based contrast agents (GBCA) safety has recently been revived due to the evidence of hyperintensities observed in the dentate nucleus (DN) and globus pallidus (GP) in the brain of patients with normal kidney function. Several preclinical studies have been conducted to understanding how the use of GBCAs can promote the gadolinium deposition in the brain. Here, we evaluate the impact of chronic cerebral hypoperfusion on gadolinium presence. METHODS: T1 hyperintensities and BBB integrity were evaluated by MRI in chronically hypoperfused and healthy rats injected with either gadodiamide or hypertonic saline. Additionally, the assessment of glucose metabolism by PET imaging and the gadolinium content by ICP-MS was performed after the last MR scan. RESULTS: Chronically hypoperfused rats displayed a greater MRI T1w signal in the DCN and hippocampus compared to Sham-operated animals, suggesting gadolinium accumulation. Dynamic contrast-enhanced (DCE) MRI assessment of BBB permeability revealed loss of integrity (high Ktrans) after rat injury in the dentate nuclei and hippocampus. Ex vivo tissue analysis showed greater gadolinium retention in the cerebellum and subcortical regions, supporting the imaging finding. FDG-PET imaging of the cerebellum did not reveal abnormal uptake in the DCN after chronic cerebral hypoperfusion. CONCLUSION: Higher signal intensity followed by higher Gd concentration observed in DCN and hippocampus of animals subjected to cerebral injury can be associated with an increase in BBB permeability due to the applied vascular dementia animal model. Nonetheless, no glucose metabolism abnormalities were detected in chronically hypoperfused cerebellum.

Photoacoustic imaging of integrin-overexpressing tumors using a novel ICG-based contrast agent in mice.

PhotoAcoustic Imaging (PAI) is a biomedical imaging modality currently under evaluation in preclinical and clinical settings. In this work, ICG is coupled to an integrin binding vector (ICG-RGD) to combine the good photoacoustic properties of ICG and the favourable αvß3-binding capabilities of a small RGD cyclic peptidomimetic. ICG-RGD is characterized in terms of physicochemical properties, biodistribution and imaging performance. Tumor uptake was assessed in subcutaneous xenograft mouse models of human glioblastoma (U-87MG, high αvß3 expression) and epidermoid carcinoma (A431, low αvß3 expression). ICG and ICG-RGD showed high PA signal in tumors already after 15 min post-injection. At later time points the signal of ICG rapidly decreased, while ICG-RGD showed sustained uptake in U-87MG but not in A431 tumors, likely due to the integrin-mediated retention of the probe. In conclusion, ICG-RGD is a novel targeted contrast agents for PAI with superior biodistribution, tumor uptake properties and diagnostic value compared to ICG.

Striatum and entorhinal cortex atrophy in AD mouse models: MRI comprehensive analysis.

Alzheimer's disease is experimentally modeled in transgenic (Tg) mice overexpressing mutated forms of the human amyloid precursor protein either alone or combined with mutated presenilins and tau. In the present study, we developed a systematic approach to compare double (TASTPM) and triple (APP/PS2/Tau) Tg mice by serial magnetic resonance imaging and spectroscopy analysis from 4 to 26 months of age to define homologous biomarkers between mice and humans. Hippocampal atrophy was found in Tg mice compared with WT. In APP/PS2/Tau the effect was age-dependent, whereas in TASTPM it was detectable from the first investigated time point. Importantly, both mice displayed an age-related entorhinal cortex thinning and robust striatal atrophy, the latter associated with a significant loss of synaptophysin. Hippocampal magnetic resonance spectroscopy revealed lower glutamate levels in both Tg mice and a selective myo-inositol increase in TASTPM. This noninvasive magnetic resonance imaging analysis, revealed common biomarkers between humans and mice, and could, thus, be promoted as a fully translational tool to be adopted in the preclinical investigation of therapeutic approaches.

Paclitaxel enhances therapeutic efficacy of the F8-IL2 immunocytokine to EDA-fibronectin-positive metastatic human melanoma xenografts.

The selective delivery of bioactive agents to tumors reduces toxicity and enhances the efficacy of anticancer therapies. In this study, we show that the antibody F8, which recognizes perivascular and stromal EDA-fibronectin (EDA-Fn), when conjugated to interleukin-2 (F8-IL2) can effectively inhibit the growth of EDA-Fn-expressing melanomas in combination with paclitaxel. We obtained curative effects with paclitaxel administered before the immunocytokine. Coadministration of paclitaxel increased the uptake of F8 in xenografted melanomas, enhancing tumor perfusion and permeability. Paclitaxel also boosted the recruitment of F8-IL2-induced natural killer (NK) cells to the tumor, suggesting a host response as part of the observed therapeutic benefit. In support of this likelihood, NK cell depletion impaired the antitumor effect of paclitaxel plus F8-IL2. Importantly, this combination reduced both the tumor burden and the number of pulmonary metastatic nodules. The combination did not cause cumulative toxicity. Together, our findings offer a preclinical proof that by acting on the tumor stroma paclitaxel potentiates the antitumor activity elicited by a targeted delivery of IL2, thereby supporting the use of immunochemotherapy in the treatment of metastatic melanoma.

E-3810 is a potent dual inhibitor of VEGFR and FGFR that exerts antitumor activity in multiple preclinical models.

Tumor angiogenesis is a degenerate process regulated by a complex network of proangiogenic factors. Existing antiangiogenic drugs used in clinic are characterized by selectivity for specific factors. Antiangiogenic properties might be improved in drugs that target multiple factors and thereby address the inherent mechanistic degeneracy in angiogenesis. Vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) family members and their cognate receptors are key players in promoting tumor angiogenesis. Here we report the pharmacologic profile of E-3810, a novel dual inhibitor of the VEGF and FGF receptors. E-3810 potently and selectively inhibited VEGF receptor (VEGFR)-1, -2, and -3 and FGF receptor (FGFR)-1 and -2 kinases in the nanomolar range. Ligand-dependent phosphorylation of VEGFR-2 and FGFR-1 was suppressed along with human vascular endothelial cell growth at nanomolar concentrations. In contrast, E-3810 lacked cytotoxic effects on cancer cell lines under millimolar concentrations. In a variety of tumor xenograft models, including early- or late-stage subcutaneous and orthotopic models, E-3810 exhibited striking antitumor properties at well-tolerated oral doses administered daily. We found that E-3810 remained active in tumors rendered nonresponsive to the general kinase inhibitor sunitinib resulting from a previous cycle of sunitinib treatment. In Matrigel plug assays performed in nude mice, E-3810 inhibited basic FGF-induced angiogenesis and reduced blood vessel density as assessed by histologic analysis. Dynamic contrast-enhanced magnetic resonance imaging analysis confirmed that E-3810 reduced the distribution of angiogenesis-sensitive contrast agents after only 5 days of treatment. Taken together, our findings identify E-3810 as a potent antiangiogenic small molecule with a favorable pharmacokinetic profile and broad spectrum antitumor activity, providing a strong rationale for its clinical evaluation.