Automatic extraction of imaging observation and assessment categories from breast magnetic resonance imaging reports with natural language processing.

BACKGROUND: Structured reports are not widely used and thus most reports exist in the form of free text. The process of data extraction by experts is time-consuming and error-prone, whereas data extraction by natural language processing (NLP) is a potential solution that could improve diagnosis efficiency and accuracy. The purpose of this study was to evaluate an NLP program that determines American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) descriptors and final assessment categories from breast magnetic resonance imaging (MRI) reports. METHODS: This cross-sectional study involved 2330 breast MRI reports in the electronic medical record from 2009 to 2017. We used 1635 reports for the creation of a revised BI-RADS MRI lexicon and synonyms lists as well as the iterative development of an NLP system. The remaining 695 reports that were not used for developing the system were used as an independent test set for the final evaluation of the NLP system. The recall and precision of an NLP algorithm to detect the revised BI-RADS MRI descriptors and BI-RADS categories from the free-text reports were evaluated against a standard reference of manual human review. RESULTS: There was a high level of agreement between two manual reviewers, with a κ value of 0.95. For all breast imaging reports, the NLP algorithm demonstrated a recall of 78.5% and a precision of 86.1% for correct identification of the revised BI-RADS MRI descriptors and the BI-RADS categories. NLP generated the total results in <1 s, whereas the manual reviewers averaged 3.38 and 3.23 min per report, respectively. CONCLUSIONS: The NLP algorithm demonstrates high recall and precision for information extraction from free-text reports. This approach will help to narrow the gap between unstructured report text and structured data, which is needed in decision support and other applications.

The applied research of simultaneous image acquisition of T2-weighted imaging (T2WI) and diffusion-weighted imaging (DWI) in the assessment of patients with prostate cancer.

We aimed to evaluate the feasibility of simultaneous image acquisition of multiple instantaneous switchable scan (MISS) for prostate magnetic resonance imaging (MRI) on 3T. Fifty-three patients were scanned with MRI due to suspected prostate cancer. Twenty-eight of them got histological results. First, two readers assessed the structure delineation and image quality based on images of conventional T2-weighted imaging (T2WI) and diffusion-weighted imaging (DWI) (CTD). Second, two readers identified the index lesion together, and then, reader one evaluated the contrast of index lesion on T2WI and signal ratio on apparent diffusion coefficient map. Third, they assigned Prostate Imaging Reporting and Data System (PI-RADS) score in consensus for the index lesion. After 4 weeks, the images of MISS were reviewed by the same readers following the same process. Finally, two readers gave preference for image interpretation, respectively. Kappa coefficient, Wilcoxon signed-rank test, paired-sample t-test, Bland-Altman analysis, and receiver operating characteristic (ROC) analysis were used for statistical analysis. The acquisition time of CTD was 6 min and 10 s, while the acquisition time of MISS was 4 min and 30 s. Interobserver agreements for image evaluation were κ = 0.65 and κ = 0.80 for CTD and MISS, respectively. MISS-T2WI showed better delineation for seminal vesicles than CTD-T2WI (reader 1: P < 0.001, reader 2: P = 0.001). The index lesion demonstrated higher contrast in MISS-T2WI (P < 0.001). The PI-RADS scores based on CTD and MISS exhibited high ability in predicting clinically significant cancer (area under curve [AUC] = 0.828 vs 0.854). Readers preferred to use MISS in 41.5%-47.2% of cases. MISS showed comparable performance to conventional technique with less acquisition time.

[Particle Size Distribution and Human Health Risk Assessment of Heavy Metals in Atmospheric Particles from Beijing and Xinxiang During Summer].

Under a condition of good air quality (AQI:55-90, PM10:37-97 µg·m-3, PM2.5:17-76 µg·m-3), six groups of 54 samples were collected using an Andersen cascade impactor from both the indoor and outdoor stations in Beijing and Xinxiang from June to August in 2016. The samples were digested by microwave digestion, and nine heavy metal elements (Pb, Cr, Ni, Cu, Zn, As, Cd, Mn, and Co) in the atmospheric particles were determined with an inductively coupled plasma source mass spectrometer (ICP-MS). The results showed that the enrichment index (0-3) of most elements were low in both cities except for Cd[15.0 (Beijing) and 8.47 (Xinxiang)]. Cr, Co, Cu, and Mn in the atmospheric particles from Beijing park, Cd, Pb, and Mn in the atmospheric particles from the Beijing office, Cr, Co, Ni, and As in the atmospheric particles from Xinxiang park, and all nine heavy metal elements in the atmospheric particles from roads in both cities were found to be more concentrated in the coarse fractions; however, Pb, Zn, Cd, Ni, and As in the atmospheric particles from Beijing park, Co, Zn, Ni, Cr, As, and Cu in the atmospheric particles from the Beijing office, Pb, Zn, Cd, Cu, and Mn in the atmospheric particles from Xinxiang park, and all nine metal elements in the atmospheric particles from the Beijing office showed the opposite pattern. The result of a human health risk assessment indicated that the carcinogenic risk of the five carcinogenic elements were all less than 10-4, but a lower potential cancer risk would also occur under long term exposure. For the four non-carcinogenic elements (Pb, Zn, Mn, and Cu), the non-carcinogenic health risk values of Pb, Zn, Mn, and Cu in the atmospheric particulates in Beijing were all far less than 1, which means the corresponding non-carcinogenic risk was negligible; and, except for Mn, there was no obvious non-carcinogenic risk from Pb, Zn, and Cu in the atmospheric particles of Xinxiang.

Controlled attenuation parameter for non-invasive assessment of hepatic steatosis in Chinese patients.

AIM: To evaluate the performance of a novel non-invasive controlled attenuation parameter (CAP) to assess liver steatosis. METHODS: This was a multi-center prospective cohort study. Consecutive patients (aged ≥ 18 years) who had undergone percutaneous liver biopsy and CAP measurement were recruited from three Chinese liver centers. Steatosis was categorized as S0: < 5%; S1: 5%-33%; S2: 34%-66%; or S3: ≥ 67%, according to the nonalcoholic fatty liver disease (NAFLD) activity score. The FibroScan(®) 502 equipped with the M probe (Echosens, Paris, France) was used to capture both CAP and liver stiffness measurement values simultaneously. Receiver operating characteristic curves were plotted, and the areas under the curves were calculated to determine the diagnostic efficacy. The accuracy of the CAP values at the optimal thresholds was defined by maximizing the sum of sensitivity and specificity (maximum Youden index). RESULTS: A total of 152 patients were recruited, including 52 (34.2%) patients with NAFLD and 100 (65.8%) with chronic hepatitis B (CHB) virus infection. After adjustment, the steatosis grade (OR = 37.12; 95%CI: 21.63-52.60, P < 0.001) and body mass index (BMI, OR = 6.20; 95%CI: 2.92-9.48, P < 0.001) were found independently associated with CAP by multivariate linear regression analysis. CAP was not influenced by inflammation, fibrosis or aetiology. The median CAP values and interquartile ranges among patients with S0, S1, S2 and S3 steatosis were 211 (181-240) dB/m, 270 (253-305) dB/m, 330 (302-360) dB/m, and 346 (313-363) dB/m, respectively. The cut-offs for the CAP values in all patients with steatosis ≥ 5%, ≥ 34% and ≥ 67% were 253 dB/m, 285 dB/m and 310 dB/m, respectively. The areas under the curves were 0.92, 0.92 and 0.88 for steatosis ≥ 5%, ≥ 34% and ≥ 67%, respectively. No significant differences were found in the CAP values between the NAFLD group and the CHB group in each steatosis grade. CONCLUSION: CAP appears to be a promising tool for the non-invasive detection and quantification of hepatic steatosis, but is limited by BMI.

[Assessment of magnetic resonance imaging in preoperative staging of endometrial carcinoma].

OBJECTIVE: To evaluate the usefulness of T2-weighed magnetic resonance images (MRI) in preoperative assessment of the depth of myometrial invasion and the staging of patients with endometrial carcinoma. METHODS: 33 cases of endometrial carcinoma were examined by MRI before operation, and the pathological findings were compared with MRI findings. RESULTS: The overall accuracy of T2-weighed magnetic resonance images in staging endometrial carcinoma was 84.85% (28/33). The accuracy for distinguishing stages I a, I b and I c was 84.62%(22/26). And the accuracy for localization of myometrial invasion (stage I c between stages I a + I b) was 96.15%(25/26). All the 3 cases with cervical invasions confirmed by pathological findings were accurately assessed by MRI. CONCLUSION: MRI is useful in assessing the depth of myometrial invasion and accurate staging in patients with endometrial carcinoma.