AI-Assisted Summarization of Radiologic Reports: Evaluating GPT3davinci, BARTcnn, LongT5booksum, LEDbooksum, LEDlegal, and LEDclinical.

BACKGROUND AND PURPOSE: The review of clinical reports is an essential part of monitoring disease progression. Synthesizing multiple imaging reports is also important for clinical decisions. It is critical to aggregate information quickly and accurately. Machine learning natural language processing (NLP) models hold promise to address an unmet need for report summarization. MATERIALS AND METHODS: We evaluated NLP methods to summarize longitudinal aneurysm reports. A total of 137 clinical reports and 100 PubMed case reports were used in this study. Models were 1) compared against expert-generated summary using longitudinal imaging notes collected in our institute and 2) compared using publicly accessible PubMed case reports. Five AI models were used to summarize the clinical reports, and a sixth model, the online GPT3davinci NLP large language model (LLM), was added for the summarization of PubMed case reports. We assessed the summary quality through comparison with expert summaries using quantitative metrics and quality reviews by experts. RESULTS: In clinical summarization, BARTcnn had the best performance (BERTscore = 0.8371), followed by LongT5Booksum and LEDlegal. In the analysis using PubMed case reports, GPT3davinci demonstrated the best performance, followed by models BARTcnn and then LEDbooksum (BERTscore = 0.894, 0.872, and 0.867, respectively). CONCLUSIONS: AI NLP summarization models demonstrated great potential in summarizing longitudinal aneurysm reports, though none yet reached the level of quality for clinical usage. We found the online GPT LLM outperformed the others; however, the BARTcnn model is potentially more useful because it can be implemented on-site. Future work to improve summarization, address other types of neuroimaging reports, and develop structured reports may allow NLP models to ease clinical workflow.

Hierarchical integration of DNA nanostructures and NanoGold onto a microchip facilitates covalent chemistry-mediated purification of circulating tumor cells in head and neck squamous cell carcinoma.

It is well-established that the combined use of nanostructured substrates and immunoaffinity agents can enhance the cell-capture performance of the substrates, thus offering a practical solution to effectively capture circulating tumor cells (CTCs) in peripheral blood. Developing along this strategy, this study first demonstrated a top-down approach for the fabrication of tetrahedral DNA nanostructure (TDN)-NanoGold substrates through the hierarchical integration of three functional constituents at various length-scales: a macroscale glass slide, sub-microscale self-organized NanoGold, and nanoscale self-assembled TDN. The TDN-NanoGold substrates were then assembled with microfluidic chaotic mixers to give TDN-NanoGold Click Chips. In conjunction with the use of copper (Cu)-catalyzed azide-alkyne cycloaddition (CuAAC)-mediated CTC capture and restriction enzyme-triggered CTC release, TDN-NanoGold Click Chips allow for effective enumeration and purification of CTCs with intact cell morphologies and preserved molecular integrity. To evaluate the clinical utility of TDN-NanoGold Click Chips, we used these devices to isolate and purify CTCs from patients with human papillomavirus (HPV)-positive (+) head and neck squamous cell carcinoma (HNSCC). The purified HPV(+) HNSCC CTCs were then subjected to RT-ddPCR testing, allowing for detection of E6/E7 oncogenes, the characteristic molecular signatures of HPV(+) HNSCC. We found that the resulting HPV(+) HNSCC CTC counts and E6/E7 transcript copy numbers are correlated with the treatment responses in the patients, suggesting the potential clinical utility of TDN-NanoGold Click Chips for non-invasive diagnostic applications of HPV(+) HNSCC.

Prostate cancer extracellular vesicle digital scoring assay - a rapid noninvasive approach for quantification of disease-relevant mRNAs.

Optimizing outcomes in prostate cancer (PCa) requires precision in characterization of disease status. This effort was directed at developing a PCa extracellular vesicle (EV) Digital Scoring Assay (DSA) for detecting metastasis and monitoring progression of PCa. PCa EV DSA is comprised of an EV purification device (i.e., EV Click Chip) and reverse-transcription droplet digital PCR that quantifies 11 PCa-relevant mRNA in purified PCa-derived EVs. A Met score was computed for each plasma sample based on the expression of the 11-gene panel using the weighted Z score method. Under optimized conditions, the EV Click Chips outperformed the ultracentrifugation or precipitation method of purifying PCa-derived EVs from artificial plasma samples. Using PCa EV DSA, the Met score distinguished metastatic (n = 20) from localized PCa (n = 20) with an area under the receiver operating characteristic curve of 0.88 (95% CI:0.78-0.98). Furthermore, longitudinal analysis of three PCa patients showed the dynamics of the Met scores reflected clinical behavior even when disease was undetectable by imaging. Overall, a sensitive PCa EV DSA was developed to identify metastatic PCa and reveal dynamic disease states noninvasively. This assay may complement current imaging tools and blood-based tests for timely detection of metastatic progression that can improve care for PCa patients.

HCC EV ECG score: An extracellular vesicle-based protein assay for detection of early-stage hepatocellular carcinoma.

BACKGROUND AND AIMS: The sensitivity of current surveillance methods for detecting early-stage hepatocellular carcinoma (HCC) is suboptimal. Extracellular vesicles (EVs) are promising circulating biomarkers for early cancer detection. In this study, we aim to develop an HCC EV-based surface protein assay for early detection of HCC. APPROACH AND RESULTS: Tissue microarray was used to evaluate four potential HCC-associated protein markers. An HCC EV surface protein assay, composed of covalent chemistry-mediated HCC EV purification and real-time immuno-polymerase chain reaction readouts, was developed and optimized for quantifying subpopulations of EVs. An HCC EV ECG score, calculated from the readouts of three HCC EV subpopulations ( E pCAM + CD63 + , C D147 + CD63 + , and G PC3 + CD63 + HCC EVs), was established for detecting early-stage HCC. A phase 2 biomarker study was conducted to evaluate the performance of ECG score in a training cohort ( n  = 106) and an independent validation cohort ( n  = 72).Overall, 99.7% of tissue microarray stained positive for at least one of the four HCC-associated protein markers (EpCAM, CD147, GPC3, and ASGPR1) that were subsequently validated in HCC EVs. In the training cohort, HCC EV ECG score demonstrated an area under the receiver operating curve (AUROC) of 0.95 (95% confidence interval [CI], 0.90-0.99) for distinguishing early-stage HCC from cirrhosis with a sensitivity of 91% and a specificity of 90%. The AUROCs of the HCC EV ECG score remained excellent in the validation cohort (0.93; 95% CI, 0.87-0.99) and in the subgroups by etiology (viral: 0.95; 95% CI, 0.90-1.00; nonviral: 0.94; 95% CI, 0.88-0.99). CONCLUSION: HCC EV ECG score demonstrated great potential for detecting early-stage HCC. It could augment current surveillance methods and improve patients' outcomes.

Drivers and Distribution of Henipavirus-Induced Syncytia: What Do We Know?

Syncytium formation, i.e., cell-cell fusion resulting in the formation of multinucleated cells, is a hallmark of infection by paramyxoviruses and other pathogenic viruses. This natural mechanism has historically been a diagnostic marker for paramyxovirus infection in vivo and is now widely used for the study of virus-induced membrane fusion in vitro. However, the role of syncytium formation in within-host dissemination and pathogenicity of viruses remains poorly understood. The diversity of henipaviruses and their wide host range and tissue tropism make them particularly appropriate models with which to characterize the drivers of syncytium formation and the implications for virus fitness and pathogenicity. Based on the henipavirus literature, we summarized current knowledge on the mechanisms driving syncytium formation, mostly acquired from in vitro studies, and on the in vivo distribution of syncytia. While these data suggest that syncytium formation widely occurs across henipaviruses, hosts, and tissues, we identified important data gaps that undermined our understanding of the role of syncytium formation in virus pathogenesis. Based on these observations, we propose solutions of varying complexity to fill these data gaps, from better practices in data archiving and publication for in vivo studies, to experimental approaches in vitro.

High-throughput screening for multi-class veterinary drug residues in animal muscle using liquid chromatography/tandem mass spectrometry with on-line solid-phase extraction.

A rapid qualitative method using on-line column-switching liquid chromatography/tandem mass spectrometry (LC/MS/MS) was developed and validated for screening 13 target veterinary drugs: four macrolides - erythromycin A, josamycin (leucomycin A3), kitasamycin (leucomycin A5), and tylosin A; six (fluoro)quinolones - ciprofloxacin, danofloxacin, enrofloxacin, flumequine, oxolinic acid, and sarafloxacin; and lincomycin, virginiamycin M1, and trimethoprim in different animal muscles. Clindamycin, norfloxacin, nalidixic acid, oleandomycin, ormetoprim, and roxithromycin were used as the internal standards. After simple deproteination and analyte extraction of muscle samples using acetonitrile, the supernatant was subjected to on-line cleanup and direct analysis by LC/MS/MS. On-line cleanup with an extraction cartridge packed with hydrophilic-hydrophobic polymer sorbent followed by fast LC using a short C18 column resulted in a total analysis cycle of 6 min for 19 drugs. This screening method considerably reduced the time and the cost for the quantitative and confirmatory analyses. The application of a control point approach was also introduced and explained.

Determination and on-line clean-up of (fluoro)quinolones in bovine milk using column-switching liquid chromatography fluorescence detection.

A simple, cost-effective, and high throughput method using on-line column-switching liquid chromatography fluorescence detection was developed and validated for analysing five (fluoro)quinolones (FQs)--enrofloxacin (ENRO), ciprofloxacin (CIPR), sarafloxacin (SARA), oxolinic acid (OXOL), and flumequine (FLUM) in bovine milk. Norfloxacin (NORF) and nalixidic acid (NALI) were used as internal standards. After simple deproteination of milk sample with 5% (w/v) metaphosphoric acid, the supernatant was subject to on-line column clean-up and direct analysis by LC-FLD. The extraction cartridge was prepared in-house by slurry packing with hydrophilic-lipophilic polymer sorbent. The accuracy of measurement for each (fluoro)quinolone at different maximum residue limits (MRL) was 101-103% (ENRO), 92.8-97.4% (CIPR), 89.8-92.8% (SARA), 116-121% (OXOL), and 81.3-85.5% (FLUM), whilst the precision was 2.9-6.1% (ENRO), 2.5-5.1% (CIPR), 2.3-5.0% (SARA), 3.1-5.9% (OXOL), and 5.6-6.5% (FLUM). The decision limits, detection capabilities, specificity and analytes stability during storage were also investigated.