Habib Zaidi

Objective. Radiomics analysis in positron emission tomography (PET) offers a high-dimensional representation of tissue characteristics, yet its reliability is limited by sensitivity to acquisition perturbations, particularly dose reduction. This study aims to introduce a nonlinear stability phenotyping framework to characterize dose–response behavior of multi-organ radiomics and evaluate their potential relevance for future digital twin-oriented applications. Approach. A multi-center dataset of 550 total-body PET scans was analyzed. Dose reduction was simulated using list-mode subsampling (dose reduction factor − DRF = 2–100). A total of 107 radiomics features were extracted per organ. Stability was quantified using intraclass correlation coefficient and summarized as stability fractions. Nonlinear models were fitted to organ- and feature-level trajectories. Stability phenotypes and critical thresholds were derived, along with organ-level instability metrics. Main results. Radiomics stability exhibited strongly nonlinear, organ-dependent behavior. Feature-level analysis identified frequent sigmoid-associated collapse patterns with high descriptive model fits ( R ² > 0.90). The liver showed the highest instability burden (30.8% unstable features), followed by esophagus and kidneys (18%–20%). Texture features were the primary source of instability, whereas shape features remained robust. Organ-level trajectories demonstrated heterogeneous responses, with some organs maintaining high stability (0.80–0.95) across DRFs, while others showed sharp declines (down to 0.64). Linear (stable) behavior dominated at the organ level, though masking underlying nonlinear feature dynamics. Significance. Radiomics stability in total-body PET follows nonlinear, phenotype-driven dynamics with identifiable tipping points and organ-specific sensitivity. The proposed framework provides a behavior-based approach for evaluating radiomics robustness and supports the selection of reliable features and organs for robust quantitative imaging and may inform future digital twin-oriented applications.

Artificial intelligence (AI) is transforming cancer management and theranostics by improving the accuracy, efficiency, and personalization of diagnostic and therapeutic workflows. Routine and accurate clinical implementation of theranostics remains limited by complex dosimetry procedures, demanding imaging protocols, and challenges in quantitative image analysis. This review critically evaluates the role of AI throughout the cancer theranostic workflow, with a focus on its potential to address the key clinical and technical challenges that continue to limit the routine implementation of personalized radiopharmaceutical therapy. It examines the current maturity of AI applications, their readiness for clinical translation, and the future prospects. Recent advances in machine learning and deep learning have enabled automated image interpretation, enhanced quantitative imaging, accelerated acquisition protocols, single-time-point dosimetry, and supported radiomics and multi-omics analyses. Emerging concepts, such as theranostic digital twins, physics- and biology-informed neural networks, and explainable AI are also discussed as future directions for precision medicine. Despite substantial progress, challenges related to data quality, interpretability, ethics, privacy, standardization, and clinical validation continue to hinder widespread clinical adoption. Nevertheless, AI-driven technologies are expected to play a central role in advancing personalized radiopharmaceutical therapy and facilitating routine dosimetry-guided treatment in clinical practice.

Brain single-photon emission computed tomography (SPECT) imaging using I-123 DaTSCAN is an effective tool for the diagnosis and follow-up of Parkinson disease. Reducing acquisition time decreases the likelihood of patient motion, improves patient comfort, and increases scanner throughput. However, shorter acquisition times often result in degraded image quality with reduced diagnostic value. This study aimed to use deep learning (DL) approaches to enhance the image quality of accelerated SPECT DaTSCAN acquisitions. A total of 85 patients were scanned on a CZT-based GE StarGuide scanner with a standard acquisition time of 15 min. To simulate fast acquisition protocols, raw list-mode data were retrospectively undersampled to represent 20%, 25%, and 50% of the original acquisition time and then reconstructed using the clinical routine protocol. A modified U-Net DL architecture incorporating attention and transformer mechanisms was trained to convert the low-count images into full-time equivalents. The output images were compared to full-time reference images using subjective image quality assessment by nuclear medicine physicians as well as region-wise and voxel-wise quantitative metrics on the test set. Both voxel-wise and region-wise metrics showed significant improvement when using the DL models for all fast acquisition times. Subjective image quality evaluations confirmed that the DL-predicted images were noticeably superior to the original low-count reconstructions in terms of perceived image quality and diagnostic confidence. The mean absolute percentage error (MAPE) was reduced from ~ 30 to ~ 18% for fast 1/5 acquisition time images and from ~ 26 to ~ 16% for 1/4 acquisition time images. Our acquisition time-specific deep learning models effectively restore information loss in fast SPECT acquisitions and reduce image noise, without compromising diagnostic value. The models offer a promising approach for faster, patient-friendly brain SPECT imaging without sacrificing image quality or quantitative accuracy.

Objective . Currently, [ ²²³ Ra]RaCl ₂ alpha-radiation radiopharmaceutical therapy ( α RPT) has been clinically approved for treating patients with metastatic castration-resistant prostate cancer (mCRPC) by employing short-range alpha-emitters for targeting bone metastasis (BM). The aim of this prospective study was to estimate the absorbed dose (AD) delivered to the target tissues post [ ²²³ Ra]RaCl ₂ α RPT in patients with mCRPC and symptomatic BMs by means of quantitative single-photon emission computed tomography/computed tomography (SPECT/CT) imaging. Approach . Four patients with mCRPC were administered [ ²²³ Ra]RaCl ₂ by injection in an activity of about 55 kBq kg ⁻¹ intravenously on a four-week schedule for up to six cycles. The dosimetry study was performed after the first treatment cycle. Pre-treatment assessment was done using [ ⁶⁸ Ga]Ga-prostate-specific membrane antigen PET/CT, [ ^99m Tc]Tc-MDP bone scan, and laboratory measurements. The SPECT/CT images were obtained after 3–6 h, 18–30 h, and 163–242 h of post-administration. The calibration phantom method was used to quantify the images, and volumes of interests delineation was performed for trabecular bone, cortical bone, bone marrow, intestines, liver, and kidneys. A tri-exponential model was applied to calculate time-integrated activity coefficients. The ADs were computed by Organ Level INternal Dose Assessment/EXponential Modeling software. Main results . There were rapid excretions from the liver and kidneys, while the bone compartments had high accumulation of [ ²²³ Ra]RaCl ₂ because of its bone-seeking effect. Large interpatient variability was noticed, especially in the intestine. High values of AD coefficients (ADCs) were found in osteogenic cells (184 mGy MBq ⁻¹ ), followed by red marrow (38.10 mGy MBq ⁻¹ ), kidneys (58 mGy MBq ⁻¹ ), and the liver (26.2 mGy MBq ⁻¹ ). The total body ADC was 7.68 mGy MBq ⁻¹ , showing selective bone targeting and low radiation ADs to soft tissue. Quantitative analysis showed the bone uptake, gastrointestinal transit, rectosigmoid accumulation, and fecal clearance. Significance . In this paper, we performed the first quantitative Ra-223 SPECT/CT imaging in mCRPC patients with BMs, providing the possibility of carrying out patient-specific dosimetry at multiple time points. The high AD for endosteal bones shows good targeting of bone and low radiation off-target in this patient group. Individualized treatment can increase the efficacy of the therapy while decreasing toxicity. Further research needs to be done on larger populations to perform more AD escalation studies.

This work evaluated the performance characteristics of the Biograph Vision.X PET/CT system according to the NEMA NU 2-2018 standards. This novel system is based on the Biograph Vision 600 with an axial field-of-view of 26.1 cm, consisting of 8 rings of 38 detector blocks using the same detector modules combining 5 × 5 array of 3.2 × 3.2 × 20 mm lutetium-oxyorthosilicate crystals with 16 × 16 mm array of silicon photomultiplier–based photodetectors. The Biograph Vision.X scanner is a newly released time-of-flight whole-body hybrid PET/CT imaging system featuring exclusive proprietary high-performance detector electronic assembly (DEA) technology with optimized digital electronics and digital processing along the entire imaging chain to achieve fast coincidence time resolution PET. In addition, examples of 3D Hoffman phantom and clinical 18F-FDG brain and whole-body studies were compared with scans performed sequentially after the Biograph Vision 600 PET/CT scanner. The Biograph Vision.X had a radial, tangential and axial spatial resolution in terms of full-width at half-maximum of 3.4, 3.5 and 3.5 mm, respectively, at a 1 cm offset from the center of the field-of-view. The sensitivity was 19.2 cps/kBq at the center and 10 cm off-center, while the peak noise-equivalent count-rate (NECR) was 291 kcps at 27.2 kBq/mL. The TOF resolution varied from 178.7 to 182.0 ps as the count rate increased up to the peak NECR. Generally, the image contrast observed on the NEMA image quality phantom ranged from 74.1% to 91.8%. The Biograph Vision.X is a competitive newly introduced device outperforming its predecessor, the Biograph Vision 600, by reaching the fastest TOF performance in the industry, thus achieving 7.5 TOF gain improvement compared to 6.2 on the Vision 600.