Empowering PET: harnessing deep learning for improved clinical insight.

This review aims to take a journey into the transformative impact of artificial intelligence (AI) on positron emission tomography (PET) imaging. To this scope, a broad overview of AI applications in the field of nuclear medicine and a thorough exploration of deep learning (DL) implementations in cancer diagnosis and therapy through PET imaging will be presented. We firstly describe the behind-the-scenes use of AI for image generation, including acquisition (event positioning, noise reduction though time-of-flight estimation and scatter correction), reconstruction (data-driven and model-driven approaches), restoration (supervised and unsupervised methods), and motion correction. Thereafter, we outline the integration of AI into clinical practice through the applications to segmentation, detection and classification, quantification, treatment planning, dosimetry, and radiomics/radiogenomics combined to tumour biological characteristics. Thus, this review seeks to showcase the overarching transformation of the field, ultimately leading to tangible improvements in patient treatment and response assessment. Finally, limitations and ethical considerations of the AI application to PET imaging and future directions of multimodal data mining in this discipline will be briefly discussed, including pressing challenges to the adoption of AI in molecular imaging such as the access to and interoperability of huge amount of data as well as the "black-box" problem, contributing to the ongoing dialogue on the transformative potential of AI in nuclear medicine.Relevance statementAI is rapidly revolutionising the world of medicine, including the fields of radiology and nuclear medicine. In the near future, AI will be used to support healthcare professionals. These advances will lead to improvements in diagnosis, in the assessment of response to treatment, in clinical decision making and in patient management.Key points• Applying AI has the potential to enhance the entire PET imaging pipeline.• AI may support several clinical tasks in both PET diagnosis and prognosis.• Interpreting the relationships between imaging and multiomics data will heavily rely on AI.

Diagnostic capabilities, clinical features, and longitudinal UBA1 clonal dynamics of a nationwide VEXAS cohort.

VEXAS is a prototypic hemato-inflammatory disease combining rheumatologic and hematologic disorders in a molecularly defined nosological entity. In this nationwide study, we aimed at screenshotting the current diagnostic capabilities and clinical-genomic features of VEXAS, and tracked UBA1 longitudinal clonal dynamics upon different therapeutics, including allogeneic hematopoietic cell transplant. We leveraged a collaboration between the Italian Society of Experimental Hematology and of Rheumatology and disseminated a national survey to collect clinical and molecular patient information. Overall, 13/29 centers performed UBA1 genomic testing locally, including Sanger sequencing (46%), next-generation sequencing (23%), droplet digital polymerase chain reaction (8%), or combination (23%). A total of 41 male patients were identified, majority (51%) with threonine substitutions at Met41 hotspot, followed by valine and leucine (27% and 8%). Median age at VEXAS diagnosis was 67 years. All patients displayed anemia (median hemoglobin 9.1 g/dL), with macrocytosis. Bone marrow vacuoles were observed in most cases (89%). The most common rheumatologic association was polychondritis (49%). A concomitant myelodysplastic neoplasm/syndrome (MDS) was diagnosed in 71% of patients (n = 28), chiefly exhibiting lower Revised International Prognostic Scoring System risk profiles. Karyotype was normal in all patients, except three MDS cases showing -Y, t(12;16)(q13;q24), and +8. The most frequently mutated gene was DNMT3A (n = 10), followed by TET2 (n = 3). At last follow-up, five patients died and two patients progressed to acute leukemia. Longitudinal UBA1 clonal dynamics demonstrated mutational clearance following transplant. We collected a nationwide interdisciplinary VEXAS patient cohort, characterized by heterogeneous rheumatologic manifestations and treatments used. MDS was diagnosed in 71% of cases. Patients exhibited various longitudinal UBA1 clonal dynamics.

Impact of inotuzumab ozogamicin on outcome in relapsed or refractory acute B-cell lymphoblastic leukemia patients prior to allogeneic hematopoietic stem cell transplantation and risk of sinusoidal obstruction syndrome/venous occlusive disease.

We evaluated 58 patients with relapsed or refractory (r/r) acute B-lymphoblastic leukemia (B-ALL; median age, 42.5 years; range, 16-69 years), treated with inotuzumab ozogamicin (INO) between 2016-2022 and who received an allogeneic hematopoietic stem cell transplantation (allo-HCT) consecutively. Forty-seven (81%) of the 58 patients were heavily pretreated receiving intensive chemotherapy +/- TKI, blinatumomab in 24 (41%) and allo-HCT at first-line in 11 (19%) patients. Complete remission (CR) rate prior to allo-HCT was 84%. Median follow-up was 30.5 months and median overall survival (OS) measured from start of INO was 11.2 months. Oneand 2-years OS rates were 50% (95%-CI, 38.4-56.1%) and 36.7% (95%-CI, 25.5-52.9%), respectively. Sinusoidal obstruction syndrome/venous occlusive disease (SOS/VOD) after allo-HCT occurred in 17 (29%) patients. Of those, 9 (53%) patients died due to SOS/VOD and multi-organ failure. Two had received >2 INO cycles (3 cycles, 5 cycles, n=1, each), all others ≤2 INO cycles prior to allo-HCT. Logistic regression analysis revealed conditioning with double alkylators (P=0.038) and allo-HCT during first-line therapy (P=0.050) as significant risk factors for SOS/VOD and in trend allo-HCT ≤ 60 days from last INO application (P=0.07), whereas number of INO cycles before allo-HCT and time between last INO application and allo-HCT were not significant. Relapse/progressive disease occurred in 20 (34%) patients. Of those, five (25%) patients are still alive, whereas 15 succumbed of their disease. Treatment with INO seems to be an effective approach with successful bridge-to-transplant. However, risk of SOS/VOD is high, necessitating continuous monitoring and recognition of SOS/VOD risk factors.

Object Counting via Group and Graph Attention Network.

Object counting, defined as the task of accurately predicting the number of objects in static images or videos, has recently attracted considerable interest. However, the unavoidable presence of background noise prevents counting performance from advancing further. To address this issue, we created a group and graph attention network (GGANet) for dense object counting. GGANet is an encoder-decoder architecture incorporating a group channel attention (GCA) module and a learnable graph attention (LGA) module. The GCA module groups the feature map into several subfeatures, each of which is assigned an attention factor through the identical channel attention. The LGA module views the feature map as a graph structure in which the different channels represent diverse feature vertices, and the responses between channels represent edges. The GCA and LGA modules jointly avoid the interference of irrelevant pixels and suppress the background noise. Experiments are conducted on four crowd-counting datasets, two vehicle-counting datasets, one remote-sensing counting dataset, and one few-shot object-counting dataset. Comparative results prove that the proposed abbr achieves superior counting performance.

Enhancing Fingerprint Liveness Detection Accuracy Using Deep Learning: A Comprehensive Study and Novel Approach.

Liveness detection for fingerprint impressions plays a role in the meaningful prevention of any unauthorized activity or phishing attempt. The accessibility of unique individual identification has increased the popularity of biometrics. Deep learning with computer vision has proven remarkable results in image classification, detection, and many others. The proposed methodology relies on an attention model and ResNet convolutions. Spatial attention (SA) and channel attention (CA) models were used sequentially to enhance feature learning. A three-fold sequential attention model is used along with five convolution learning layers. The method's performances have been tested across different pooling strategies, such as Max, Average, and Stochastic, over the LivDet-2021 dataset. Comparisons against different state-of-the-art variants of Convolutional Neural Networks, such as DenseNet121, VGG19, InceptionV3, and conventional ResNet50, have been carried out. In particular, tests have been aimed at assessing ResNet34 and ResNet50 models on feature extraction by further enhancing the sequential attention model. A Multilayer Perceptron (MLP) classifier used alongside a fully connected layer returns the ultimate prediction of the entire stack. Finally, the proposed method is also evaluated on feature extraction with and without attention models for ResNet and considering different pooling strategies.

Dynamics of polyclonal immuno-reconstitution after allogeneic transplant with post-transplant cyclophosphamide and letermovir.

Cytomegalovirus (CMV) reactivations are strong stimulators of immune-reconstitution (IR) in hematopoietic stem cell transplantation (HSCT) recipients. Herein, we analyzed 317 CMV-seropositive consecutive patients (n = 109 letermovir, LTV; n = 208 no-LTV), undergoing HSCT with post-transplant cyclophosphamide (PTCy) and calcineurin inhibitor- (CNI) free graft-versus-host-disease (GvHD) prophylaxis. At day+90, median CD19+/mm3 was higher in LTV-cohort: 5.5 [0;439] versus 2 [0;294], p = 0.008; median CD3+/mm3 counts were lower in LTV-cohort, with no differences in CD4+, CD8+ and NK-cells. At day+180 median CD3+, CD4+ and CD8+/mm3 values were comparable between groups. Higher CD19+/mm3 counts were observed in LTV-cohort: 62 [0; 2983] versus 42 [0; 863]. Significantly higher median NK/mm3 values were seen in LTV-cohort: 225.5 [0;763] versus 163.5 [0;1181], p = 0.0003. The impact of LTV on B-cell IR at 3 months and NK-cell levels at 6 months was retained in multivariate analysis (p < 0.01), whereas the effect on T-cells was not confirmed. Moreover, we confirmed a significant reduction of clinically-relevant CMV, and moderate-to- severe chronic GvHD in LTV-cohort. Overall, in our study the use of LTV was associated with a slight improvement of B-cell and NK-cells reconstitution, with only minor impact on T-cell subsets, giving new insights on polyclonal IR for HSCT recipients in the LTV era.

Splenic irradiation prior to allogeneic transplant conditioning in myelofibrosis: A pilot experience.

INTRODUCTION: In the era of JAK inhibitors, allogeneic stem cell transplantation (HSCT) remains the only curative treatment for patients with Myelofibrosis (MF). Splenic irradiation (SI) may be used to reduce spleen size and related symptoms. METHODS: We conducted a retrospective analysis on 14 patients with MF who underwent HSCT with SI from any donor source at our center between June 2016 and March 2021. All patients received a conditioning backbone based on treosulfan and fludarabine, with post-transplant cyclophosphamide (PTCy) and sirolimus as graft-versus-host disease (GvHD) prophylaxis. Patients received SI with 10 Gy involved-field radiotherapy in five 2-Gy fractions over the course of a week prior to the beginning of conditioning. RESULTS: At transplant all patients were transfusion-dependent and had splenomegaly (median bipolar diameter by ultrasound: 20.75 cm). Overall, 12 patients had received ruxolitinib prior to transplant. Re-evaluation of spleen dimensions was available for 13 patients: median splenic bipolar diameter after at least 3 months from transplant decreased by a median of 25%. With a median post-transplant follow-up of 25 months, 6 patients remain in CR with full-donor chimerism, 3 patients died due to NRM. Overall, 4 patients relapsed. At last follow-up, nine patients are currently alive and achieved transfusion-independence. CONCLUSIONS: In a small cohort of mostly ruxolitinib pre-treated patients, SI and treosulfan-based conditioning appeared a safe and effective tool to reduce spleen dimensions and ameliorate symptoms. Future prospective studies with adequate sample size are warranted to further investigate the usefulness and safety of this approach in MF.

Case Report: Favorable outcome of allogeneic hematopoietic stem cell transplantation in SARSCoV2 positive recipient, risk-benefit balance between infection and leukemia.

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) in SARS-CoV-2 positive candidates is usually delayed until the clinical resolution of the infection's symptoms and a negative nasopharyngeal molecular test. However, prolonged SARS-CoV-2 positivity has been frequently observed in haematological malignancies, thus representing a challenge for the timing of transplant procedures. Here, we report on the case of a 34-year-old patient with recent pauci-symptomatic COVID-19 undergoing transplant for high-risk acute B-lymphoblastic leukemia before achieving viral clearance. Shortly before their scheduled allogeneic HSCT from a matched unrelated donor, the patient developed mild Omicron BA.5 infection receiving nirmatrelvir/ritonavir with fever resolution within 72 hours. Twenty-three days after COVID-19 diagnosis, because of increasing minimal residual disease values in the context of high-risk refractory leukemia and clinical resolution of SARS-2-CoV infection with reduction of viral load at surveillance nasopharyngeal swabs, it was decided not to delay further allo-HSCT. During myelo-ablative conditioning, the nasopharyngeal SARS-CoV-2 viral load increased while the patient remained asymptomatic. Consequently, two days before the transplant, intra-muscular tixagevimab/cilgavimab 300/300 mg and a 3-day course of intravenous remdesivir were administered. During the pre-engraftment phase, veno-occlusive disease (VOD) occurred at day +13, requiring defibrotide treatment to obtain a slow but complete recovery. The post-engraftment phase was characterized by mild COVID-19 at day +23 (cough, rhino-conjunctivitis, fever) that spontaneously resolved, achieving viral clearance at day +28. At day +32, she experienced grade I acute graft-versus host disease (a-GVHD, skin grade II) treated with steroids and photo-apheresis, without further complications during follow-up until day +180. Addressing the issue of allo-HSCT timing in patients recovering from SARS-CoV-2 infection with high-risk malignant diseases is challenging because of 1] the high risk of COVID-19 clinical progression, 2] the impact of transplant delay on leukemia prognosis and 3] the occurrence of endothelial complications such as VOD, a-GVHD, and transplant associated thrombotic micro-angiopathy. Our report describes the favourable outcome of allo-HSCT in a recipient with active SARS-CoV2 infection and high-risk leukemia thanks to timely anti-SARS-CoV-2 preventive therapies and prompt management of transplant-related complications.

Schrödinger cat states of a 16-microgram mechanical oscillator.

According to quantum mechanics, a physical system can be in any linear superposition of its possible states. Although the validity of this principle is routinely validated for microscopic systems, it is still unclear why we do not observe macroscopic objects to be in superpositions of states that can be distinguished by some classical property. Here we demonstrate the preparation of a mechanical resonator in Schrödinger cat states of motion, where the ∼1017 constituent atoms are in a superposition of two opposite-phase oscillations. We control the size and phase of the superpositions and investigate their decoherence dynamics. Our results offer the possibility of exploring the boundary between the quantum and classical worlds and may find applications in continuous-variable quantum information processing and metrology with mechanical resonators.

Autoimmune manifestations in VEXAS: Opportunities for integration and pitfalls to interpretation.

VEXAS (Vacuoles, E1 enzyme, X-linked, Autoinflammatory, Somatic) is a novel entity manifesting with a multiplicity of clinical features. Somatic mutations of the UBA1 gene in hematopoietic stem cells constitute the genetic basis of VEXAS. As an X-linked disorder, most cases occur in men, classically developing symptoms during the fifth to sixth decade of life. Considering its multidisciplinary nature involving numerous branches of internal medicine, VEXAS has elicited a wide medical interest and several medical conditions have been associated with this disease. Even so, its recognition in everyday clinical practice is not necessarily straightforward. Close collaboration between different medical specialists is mandatory. Patients with VEXAS may manifest a range of features from manageable cytopenias to disabling and life-threatening autoimmune phenomena with limited responses to therapy, with the potential for progression to hematological malignancies. Diagnostic and treatment guidelines are exploratory and include a range of rheumatological and supportive care treatments. Allogeneic hematopoietic stem cell transplantation is potentially curative, but its risks are significant and its position in the treatment algorithm is yet to be defined. Herein, we present the variegated manifestations of VEXAS, provide practice criteria for diagnostic testing of UBA1, and discuss potential treatment options, including allogeneic hematopoietic stem cell transplantation, current evidence, and future directions.

The Development of an Intelligent Agent to Detect and Non-Invasively Characterize Lung Lesions on CT Scans: Ready for the "Real World"?

(1) Background: Once lung lesions are identified on CT scans, they must be characterized by assessing the risk of malignancy. Despite the promising performance of computer-aided systems, some limitations related to the study design and technical issues undermine these tools' efficiency; an "intelligent agent" to detect and non-invasively characterize lung lesions on CT scans is proposed. (2) Methods: Two main modules tackled the detection of lung nodules on CT scans and the diagnosis of each nodule into benign and malignant categories. Computer-aided detection (CADe) and computer aided-diagnosis (CADx) modules relied on deep learning techniques such as Retina U-Net and the convolutional neural network; (3) Results: Tests were conducted on one publicly available dataset and two local datasets featuring CT scans acquired with different devices to reveal deep learning performances in "real-world" clinical scenarios. The CADe module reached an accuracy rate of 78%, while the CADx's accuracy, specificity, and sensitivity stand at 80%, 73%, and 85.7%, respectively; (4) Conclusions: Two different deep learning techniques have been adapted for CADe and CADx purposes in both publicly available and private CT scan datasets. Experiments have shown adequate performance in both detection and diagnosis tasks. Nevertheless, some drawbacks still characterize the supervised learning paradigm employed in networks such as CNN and Retina U-Net in real-world clinical scenarios, with CT scans from different devices with different sensors' fingerprints and spatial resolution. Continuous reassessment of CADe and CADx's performance is needed during their implementation in clinical practice.

Sign and Human Action Detection Using Deep Learning.

Human beings usually rely on communication to express their feeling and ideas and to solve disputes among themselves. A major component required for effective communication is language. Language can occur in different forms, including written symbols, gestures, and vocalizations. It is usually essential for all of the communicating parties to be fully conversant with a common language. However, to date this has not been the case between speech-impaired people who use sign language and people who use spoken languages. A number of different studies have pointed out a significant gaps between these two groups which can limit the ease of communication. Therefore, this study aims to develop an efficient deep learning model that can be used to predict British sign language in an attempt to narrow this communication gap between speech-impaired and non-speech-impaired people in the community. Two models were developed in this research, CNN and LSTM, and their performance was evaluated using a multi-class confusion matrix. The CNN model emerged with the highest performance, attaining training and testing accuracies of 98.8% and 97.4%, respectively. In addition, the model achieved average weighted precession and recall of 97% and 96%, respectively. On the other hand, the LSTM model's performance was quite poor, with the maximum training and testing performance accuracies achieved being 49.4% and 48.7%, respectively. Our research concluded that the CNN model was the best for recognizing and determining British sign language.

Post-transplant cyclophosphamide and sirolimus based graft-versus-host disease prophylaxis after allogeneic stem cell transplantation for acute myeloid leukemia.

Post-transplant cyclophosphamide (PTCy) has emerged as a promising graft-versus-host disease (GvHD) prophylaxis in allogeneic hematopoietic stem cell transplantation (allo-HSCT). However, no studies have reported the efficacy of a GvHD prophylaxis based on PTCy with sirolimus (Sir-PTCy) in patients with acute myeloid leukemia (AML). In this retrospective study, we analyze the use of sirolimus in combination with PTCy, with or without mycophenolate mofetil (MMF), on 242 consecutive adult patients with AML undergoing a myeloablative first allo-HSCT from different donor types, in three European centers between January 2017 and December 2020. Seventy-seven (32%) patients received allo-HSCT from HLA-matched sibling donor, 101 (42%) from HLA-matched and mismatched unrelated donor, and 64 (26%) from haploidentical donor. Except for neutrophil and platelet engraftment, which was slower in the haploidentical cohort, no significant differences were observed in major transplant outcomes according to donor type in univariate and multivariate analysis. GvHD prophylaxis with Sir-PTCy, with or without MMF, is safe and effective in patients with AML undergoing myeloablative allo-HSCT, resulting in low rates of transplant-related mortality, relapse/progression, and acute and chronic GvHD in all donor settings.

Image Augmentation Techniques for Mammogram Analysis.

Research in the medical imaging field using deep learning approaches has become progressively contingent. Scientific findings reveal that supervised deep learning methods' performance heavily depends on training set size, which expert radiologists must manually annotate. The latter is quite a tiring and time-consuming task. Therefore, most of the freely accessible biomedical image datasets are small-sized. Furthermore, it is challenging to have big-sized medical image datasets due to privacy and legal issues. Consequently, not a small number of supervised deep learning models are prone to overfitting and cannot produce generalized output. One of the most popular methods to mitigate the issue above goes under the name of data augmentation. This technique helps increase training set size by utilizing various transformations and has been publicized to improve the model performance when tested on new data. This article surveyed different data augmentation techniques employed on mammogram images. The article aims to provide insights into basic and deep learning-based augmentation techniques.

Addition of a Single Low Dose of Anti T-Lymphocyte Globulin to Post-Transplant Cyclophosphamide after Allogeneic Hematopoietic Stem Cell Transplant: A Pilot Study.

Correlation between risk of graft-versus-host disease (GvHD) and CD3+ counts within the peripheral blood stem cell graft has recently been reported in the setting of post-transplant cyclophosphamide (PT-Cy). We aimed to investigate the benefit of the addition of a single dose of anti-T lymphocyte globulin (ATLG 5 mg/kg) to PT-Cy in this setting. Starting in 2019, all patients receiving PBSC transplant containing CD3+ counts above 300 × 106/kg (study group) received a post-transplant dose of ATLG in addition to standard PT-Cy. The study was designed as a real-life analysis and included all consecutive Hematopoietic Stem Cell Transplantation (HSCT) recipients according to the above-mentioned inclusion criterion (n = 21), excluding cord blood and bone marrow donors. Using a 1:2 matched-pair analysis, we compared the outcomes with a historical population who received PT-Cy only (control group). We found a delayed platelet engraftment (29% vs. 45% at 30 days, p = 0.03) and a non-significant trend toward higher risk of poor graft function (29% vs. 19%, p = 0.52). The addition of ATLG impacted long-term immune reconstitution on the CD4+ subsets, but this did not translate into higher rate of relapse or viral infection. Acute GvHD was not significantly impacted, but 1-year cumulative incidence of chronic GvHD was significantly lower in the study group (15% vs. 41%, p = 0.04). Survival outcomes were comparable. In conclusion PT-Cy and ATLG was overall safe and translated into a low rate of chronic GvHD incidence.