Ana Ribeiro

Background: Oncology practice is increasingly aiming to be patient centric. Imaging is a decisive part of the management of cancer patients and with the introduction of Digital Health Records (DHR) patients have the possibility of accessing their imaging results independently, yet the optimal way of doing so is still not clear. The introduction of Large Language Models (LLM) offers the potential to turn radiology reports into a clearer, accessible and unambiguous format and to democratise patient’s access to their own medical records. Methods: A multi-reader retrospective Service Evaluation (SE) conducted at a tertiary oncology hospital aimed to assess the capability of an LLM to generate two versions of simplified oncology imaging reports. The SE assessed Patient and Public Involvement (PPI) representatives and healthcare professionals’ (HCP) preferences using original radiology reports from two cohorts, colorectal ( n = 30) and lung ( n = 30) cancer. A Prompt-development phase created two prompts to generate the summarised (version A) and the full-length (version B) report versions. The review was performed by radiologists with 360 reads and PPI representatives with 180 reads. Results: Radiologists scores between summaries and full-length reports differed per cohort. In the lung cohort, version A was rated higher for factual correctness ( P = 0.001), completeness ( P < 0.0001), accessibility and readability ( P = 0.026), and benefit to patients ( P < 0.0001). The opposite was seen in the colorectal cohort, version B achieved consistently higher scores ( P < 0.002). When the two cohorts were combined, median scores for version A and B did not differ significantly (all P > 0.057). PPI reviews indicated that full-length reports were favoured significantly ( P < 0.0001). Qualitative results from radiologists and PPI identified incorrect statements ( n = 28), complex terminology ( n = 18), addition of confusion ( n = 10), and missing information ( n = 10). Conclusions: LLM simplified reports have the potential to improve patient accessibility in oncology imaging. PPI and HCP preferences for summarised versus full-length reports vary. Findings suggest these outputs are likely to benefit from appropriate adjustments to individual patient needs and clinical context. Reports with incorrect, confusing and missing content, highlight that LLM need improvement, ahead of potential clinical use in this setting.

Objective . Quantitative assessment of treatment response in advanced prostate cancer (APC) with bone metastases remains an unmet clinical need. Whole-body diffusion-weighted MRI (WB-DWI) provides two response biomarkers: total diffusion volume (TDV) and global apparent diffusion coefficient (gADC). However, tracking post-treatment changes of TDV and gADC from manually delineated lesions is cumbersome and increases inter-reader variability. We developed a software to automate this process. Approach . Core technologies include: (i) a weakly-supervised Residual U-Net model generating a skeleton probability map to isolate bone; (ii) a statistical framework for WB-DWI intensity normalisation, obtaining a signal-normalised b = 900 s mm ⁻² (b900) image; and (iii) a shallow convolutional neural network that processes outputs from (i) and (ii) to generate a mask of suspected bone lesions, characterised by higher b900 signal intensity due to restricted water diffusion. This mask is applied to the gADC map to extract TDV and gADC statistics. We tested the tool using expert-defined metastatic bone disease delineations on 66 datasets, assessed repeatability of imaging biomarkers ( N = 10), and compared software-based response assessment with a construct reference standard , defined as multidisciplinary consensus based on ⩾12 months of imaging, clinical, and laboratory follow-up ( N = 118). Main results . Average dice score between manual and automated delineations was 0.6 for lesions within pelvis and spine, with an average surface distance of 2 mm. Relative differences for log-transformed TDV (log-TDV) and median gADC were 8.8% and 5%, respectively. Repeatability analysis showed coefficients of variation of 4.6% for log-TDV and 3.5% for median gADC, with intraclass correlation coefficients of 0.94 or higher. The software achieved 80.5% accuracy, 84.3% sensitivity, and 85.7% specificity in assessing response to treatment. Average computation time was 90 s per scan. Significance . Our software enables reproducible TDV and gADC quantification from WB-DWI scans for monitoring metastatic bone disease response, thus providing potentially useful measurements for clinical decision-making in APC patients.

Background: Whole-Body Diffusion-Weighted Imaging (WBDWI) is an established technique for staging and evaluating treatment response in patients with multiple myeloma (MM) and advanced prostate cancer (APC). However, WBDWI scans show inter- and intra-patient intensity signal variability. This variability poses challenges in accurately quantifying bone disease, tracking changes over follow-up scans, and developing automated tools for bone lesion delineation. Here, we propose a novel automated pipeline for inter-station, inter-scan image signal standardisation on WBDWI that utilizes robust segmentation of the spinal canal through deep learning. Methods: We trained and validated a supervised 2D U-Net model to automatically delineate the spinal canal (both the spinal cord and surrounding cerebrospinal fluid, CSF) in an initial cohort of 40 patients who underwent WBDWI for treatment response evaluation (80 scans in total). Expert-validated contours were used as the target standard. The algorithm was further semi-quantitatively validated on four additional datasets (three internal, one external, 207 scans total) by comparing the distributions of average apparent diffusion coefficient (ADC) and volume of the spinal cord derived from a two-component Gaussian mixture model of segmented regions. Our pipeline subsequently standardises WBDWI signal intensity through two stages: (i) normalisation of signal between imaging stations within each patient through histogram equalisation of slices acquired on either side of the station gap, and (ii) inter-scan normalisation through histogram equalisation of the signal derived within segmented spinal canal regions. This approach was semi-quantitatively validated in all scans available to the study (N = 287). Results: The test dice score, precision, and recall of the spinal canal segmentation model were all above 0.87 when compared to manual delineation. The average ADC for the spinal cord (1.7 × 10⁻³ mm²/s) showed no significant difference from the manual contours. Furthermore, no significant differences were found between the average ADC values of the spinal cord across the additional four datasets. The signal-normalised, high-b-value images were visualised using a fixed contrast window level and demonstrated qualitatively better signal homogeneity across scans than scans that were not signal-normalised. Conclusion: Our proposed intensity signal WBDWI normalisation pipeline successfully harmonises intensity values across multi-centre cohorts. The computational time required is less than 10 s, preserving contrast-to-noise and signal-to-noise ratios in axial diffusion-weighted images. Importantly, no changes to the clinical MRI protocol are expected, and there is no need for additional reference MRI data or follow-up scans.

There is continuing debate concerning the risks of secondary malignancies from low levels of radiation exposure. The current model used for radiation protection is predicated on the assumption that even very low levels of exposure may entail risk. This has profound implications for medical procedures involving ionising radiation as radiation doses must be carefully monitored, and for diagnostic procedures are minimised as far as possible. This incurs considerable expense. The SOLLID study (ClinicalTrials.gov Identifier: NCT03580161) aims to develop the methodology to enable a large-scale epidemiological investigation of the effect of radiopharmaceutical administrations to patients undergoing diagnostic nuclear medicine procedures. Patients will undergo a series of scans in addition to that acquired as standard of care to enable the radiation doses delivered to healthy organs to be accurately calculated. Detailed analysis will be performed to determine the uncertainty in the radiation dose calculations as a function of the number and type of scans acquired. It is intended that this will inform a subsequent long-term multicentre epidemiological study that would address the question definitively. Secondary aims of the study are to evaluate the range of absorbed doses that are delivered from diagnostic nuclear medicine procedures and to use current risk models to ascertain the relative risks from these administrations.