The Royal College of Radiologists Open最新文献

Purpose

To investigate the reliability of ChatGPT in grading imaging requests using the Reason for exam Imaging Reporting and Data System (RI-RADS).

Method

In this single-center retrospective study, a total of 450 imaging referrals were included. Two human readers independently scored all requests according to RI-RADS. We created a customized RI-RADS GPT where the requests were copied and pasted as inputs, getting as an output the RI-RADS score along with the evaluation of its three subcategories. Pearson's chi-squared test was used to assess whether the distributions of data assigned by the radiologist and ChatGPT differed significantly. Inter-rater reliability for both the overall RI-RADS score and its three subcategories was assessed using Cohen's kappa (κ).

Results

RI-RADS D was the most prevalent grade assigned by humans (54% of cases), while ChatGPT more frequently assigned the RI-RADS C (33% of cases). In 2% of cases, ChatGPT assigned the wrong RI-RADS grade, based on the ratings given to the subcategories. The distributions of the RI-RADS grade and the subcategories differed statistically significantly between the radiologist and ChatGPT, apart from RI-RADS grades C and X. The reliability between the radiologist and ChatGPT in assigning RI-RADS score was very low (κ: 0.20), while the agreement between the two human readers was almost perfect (κ: 0.96).

Conclusions

ChatGPT may not be reliable for independently scoring the radiology exam requests according to RI-RADS and its subcategories. Furthermore, the low number of complete imaging referrals highlights the need for improved processes to ensure the quality of radiology requests.

Purpose

To describe the overall treatment patterns of follicular lymphoma (FL) in a single tertiary institution, with a focus on the outcomes of radiotherapy treatment in this group of patients.

Methods and materials

Retrospective analysis on 177 patients with FL registered from 1996 to 2011 and followed up to October 2017. Diagnosis was confirmed by histology and staging by clinical examination and imaging with positron emission tomography (PET) or computed tomography (CT) scan. Demographics, clinicopathological characteristics, treatment details and clinical outcomes were reviewed.

Results

Fifty-five patients (31%) had early stage and 122 (69%) had advanced stage FL. Median follow-up was 7 years and 7.5 years for early and advanced FL respectively. For advanced stage, median overall survival (OS) was 12.9 years, 5-year and 10-year OS were 77% and 63%, and 5-year and 10-yr PFS were 57% and 40% respectively. In multivariable model, moderate FLIPI, palliative RT and anti-CD20 targeted therapy were significantly associated with OS when comparing with those with high FLIPI score, no RT and no anti-CD20 targeted therapy respectively. For early stage, median OS has not reached, 5-year and 10-year OS were 79% and 74% and 5-year and 10-year progression free survival (PFS) were 64% and 52% respectively. Stage 2 non-contiguous disease had significantly poorer survival than stage 1 patients in multivariable (HR 8.19, 95%CI 1.86 – 35.98) analysis. Overall, 49 of 177 patients (27.7%) had any form of radiotherapy as part of their treatment for FL. There were no in-field alone relapses seen in all cases treated with RT. 10% in early FL had G3-4 acute toxicities with RT whilst there was no G3-4 late toxicities. In advanced FL, there were no G3-4 acute toxicities with consolidation RT and 1 G3 with salvage RT.

Conclusion

Various treatment modalities were used to treat FL. FL has effective response rates to RT both in early and advanced FL. Amongst early stage, Stage 2 non-contiguous disease tended to have poorer outcomes, indicating the need for more aggressive treatment strategies for this group of patients. Patients with advanced FL may benefit from RT. There were no in-field alone relapses seen in all cases treated with RT in the early or advanced stage. New prospective studies are warranted to shed light on improving the utility of RT for patients with FL. With advances in molecular medicine, imaging and treatment approaches, there is potential for better stratification of different groups for more localized versus more systemic treatments.

Purpose

Methods

Results

Conclusion

Rationale and objectives

Dual-energy CT (DECT) plays an important role in the evaluation of gout based on its ability to identify monosodium urate (MSU) crystals. Post-processing of DECT includes selection of a minimum Hounsfield unit (HU) threshold, currently set at 150 HU. Adjustments to this setting will affect the amount of MSU that is identified, with lower thresholds identifying more true and false positive MSU. The purpose of this study is to assess the performance of several HU thresholds.

Materials and methods

134 DECTs were reprocessed with minimum HU thresholds set to 150, 140, 130, and 120 HU. Volume of MSU detected was compared across thresholds. Subsequently, readers reviewed 3D axial images from 109 cases at 150, 130, and 120 HU thresholds. For each case, the reader made a diagnosis of gout, rated their confidence, and reported the location of MSU deposits.

Results

Mean MSU volume detected was as follows: 0.160 at 150, 0.205 at 140, 0.276 at 130, and 0.444 at 120 HU. Based on reader diagnosis, there was no significant difference in sensitivity for the diagnosis of gout between the 150 (68.8%), 130 (71.8%), and 120 (69.7%) HU thresholds (p = 0.653). Specificity was higher at 150 HU (65.4%) than at 130 (58.4%) and 120 (52.2%), (p < 0.001). Mean reader confidence was 90.1 for 150 HU, 79.9 for 130 HU, and 74.0 for 120 HU (p < 0.001).

Conclusion

Between the 150, 130, and 120 HU thresholds, 150 HU performed best overall for making the diagnosis of gout on DECT. Further research is needed to determine if there is utility for utilizing lower threshold settings for certain clinical scenarios.

Background

This study was done to assess inter and intra-rater bias in rating of cosmesis, when breast photographs were evaluated serially or randomly by a panel of six members having varying years of experience.

Methods

Cosmetic assessment was done subjectively for 175 images [of 50 unilateral breast cancer patients for whom at least 3 images were collected], that were arranged serially from baseline to follow up in chronological order termed ‘serial assessment setting’ [SAS]. For ‘random assessment setting’ [RAS], all images was randomly arranged for assessment. Objectively assessment was also done using BCCT.core. Kappa index was calculated for agreement between the RAS and SAS rating for the 3 panellists' groups and with BCCT.core.

Results

Good agreement [kappa 0.659] was found between the mean panel cosmetic scores for both SAS and RAS. Fair agreement was found when subjective RAS [k = 0.301] and SAS [k = 0.343] scores were compared with the BCCT.core, which was highest for the most experienced panellists with SAS k = 0.387 and RAS k = 0.436. Both SAS and RAS had good intra-rater reliability.

Conclusions

SAS improves the agreement with BCCT.core rating and may be used if validated in a larger cohort. The clinical experience of the panellist impacts cosmetic rating and must be considered before forming a panel.

The purpose of this study is to present the indications, rationale and techniques used in minimally invasive percutaneous biopsy of the bowel. Both efficacy and rate of complications were evaluated. A series of 63 patients over a ten-year period were retrospectively reviewed. Techniques using fluoroscopy, ultrasound and CT guidance are described.

Percutaneous biopsy of the bowel is a safe and effective method for obtaining histological samples in selected patients who are not fit for endoscopy or for other technical reasons. Radiologists should be aware of its potential as a valid alternative.