Na Liu

Na Liu


Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 December 2024 in Journal of Medical Internet Research
Verified
Journal Article
Open access
ZZZiwei Zhao
Ziwei Zhao
WZWeiyi Zhang
Weiyi Zhang
XCXiaolan Chen
Xiaolan Chen
FSFan Song
Fan Song
JGJames Gunasegaram
James Gunasegaram
NLNa Liu
Na Liu
Background: Large language models have shown remarkable efficacy in various medical research and clinical applications. However, their skills in medical image recognition and subsequent report generation or question answering (QA) remain limited. Objective: We aim to finetune a multimodal, transformer-based model for generating medical reports from slit lamp images and develop a QA system using Llama2. We term this entire process slit lamp–GPT. Methods: Our research used a dataset of 25,051 slit lamp images from 3409 participants, paired with their corresponding physician-created medical reports. We used these data, split into training, validation, and test sets, to finetune the Bootstrapping Language-Image Pre-training framework toward report generation. The generated text reports and human-posed questions were then input into Llama2 for subsequent QA. We evaluated performance using qualitative metrics (including BLEU [bilingual evaluation understudy], CIDEr [consensus-based image description evaluation], ROUGE-L [Recall-Oriented Understudy for Gisting Evaluation—Longest Common Subsequence], SPICE [Semantic Propositional Image Caption Evaluation], accuracy, sensitivity, specificity, precision, and F1-score) and the subjective assessments of two experienced ophthalmologists on a 1-3 scale (1 referring to high quality). Results: We identified 50 conditions related to diseases or postoperative complications through keyword matching in initial reports. The refined slit lamp–GPT model demonstrated BLEU scores (1-4) of 0.67, 0.66, 0.65, and 0.65, respectively, with a CIDEr score of 3.24, a ROUGE (Recall-Oriented Understudy for Gisting Evaluation) score of 0.61, and a Semantic Propositional Image Caption Evaluation score of 0.37. The most frequently identified conditions were cataracts (22.95%), age-related cataracts (22.03%), and conjunctival concretion (13.13%). Disease classification metrics demonstrated an overall accuracy of 0.82 and an F1-score of 0.64, with high accuracies (≥0.9) observed for intraocular lens, conjunctivitis, and chronic conjunctivitis, and high F1-scores (≥0.9) observed for cataract and age-related cataract. For both report generation and QA components, the two evaluating ophthalmologists reached substantial agreement, with κ scores between 0.71 and 0.84. In assessing 100 generated reports, they awarded scores of 1.36 for both completeness and correctness; 64% (64/100) were considered “entirely good,” and 93% (93/100) were “acceptable.” In the evaluation of 300 generated answers to questions, the scores were 1.33 for completeness, 1.14 for correctness, and 1.15 for possible harm, with 66.3% (199/300) rated as “entirely good” and 91.3% (274/300) as “acceptable.” Conclusions: This study introduces the slit lamp–GPT model for report generation and subsequent QA, highlighting the potential of large language models to assist ophthalmologists and patients.

23 October 2024 in Surgical Endoscopy
Verified
Journal Article
Closed access
PDF file
YXYan Xu
Yan Xu
SLSiquan Li
Siquan Li
ZYZhanhui Ye
Zhanhui Ye
XLXiue Liang
Xiue Liang
WZWeizheng Zhang
Weizheng Zhang
KCKequan Chen
Kequan Chen

4 July 2025 in Scientific Reports
Verified
Journal Article
Open access
PDF file
YXYan Xu
Yan Xu
HHHongzhen He
Hongzhen He
ZYZhanhui Ye
Zhanhui Ye
SLSiquan Li
Siquan Li
HLHu Li
Hu Li
XCXiangsheng Cai
Xiangsheng Cai
To explore the relationship between the incidence of colorectal polyps and related factors such as markers of insulin resistance (IR) , a retrospective analysis was used to collect 2716 patients who underwent physical examinations in Guangzhou Cadre and Talent Health Management Center from June 2020 to June 2023. The patients were classified as colorectal polyp group (n = 1328) and non-polyp group (n = 1388) after colonoscopy. Baseline data of colorectal polyp and non-polyp groups were compared. Non-insulin-based IR markers including Triglyceride Glucose Index (TyG), triglyceride glucose index (TyG)- Body Mass Index(BMI), triglycerides (TG)/high density lipoprotein cholesterol (HDL-C), Metabolic Score for Insulin Resistance (METS-IR), and mean arterial pressure (MAP) were calculated. The risk factors of colorectal polyps were analyzed by binary logistic regression. Receiver operating characteristic (ROC) curves and area under the curve (AUC) were used to evaluate the ability of insulin resistance markers to detect colorectal polyps. The proportion of males in colorectal polyp group was higher than that in non-polyp group. There were significant differences in TyG-BMI, TyG, TG/HDL-C, METS-IR, MAP, and other indicators between the colorectal polyp group and the non-polyp group (p < 0.05). Binary logistic regression analysis showed that elevation in age, BMI, TyG-BMI, TyG, TG/HDL-C, METS-IR, MAP and CEA were positively correlated with the risk of colorectal polyps (p < 0.05). The AUCs of insulin resistance markers and MAP were larger than CEA to detect colorectal polyps. The AUC values of models in combination of age, sex and insulin resistance markers/MAP is higher. TyG, TyG-BMI, TG/HDL-C, and METS-IR could be used as indicators to predict the occurrence of colorectal polyps. These findings provide an important clinical reference for the prevention and treatment of colorectal polyps.

27 October 2023 in JMIR Preprints
Verified
Preprint
Open access
ZZZiwei Zhao
Ziwei Zhao
WZWeiyi Zhang
Weiyi Zhang
XCXiaolan Chen
Xiaolan Chen
FSFan Song
Fan Song
JGJames Gunasegaram
James Gunasegaram
NLNa Liu
Na Liu
BACKGROUND Large language models have shown remarkable efficacy in various medical research and clinical applications. However, their skills in medical image recognition and subsequent report generation or question answering (QA) remain limited. OBJECTIVE We aim to finetune a multimodal, transformer-based model for generating medical reports from slit lamp images and develop a QA system using Llama2. We term this entire process slit lamp–GPT. METHODS Our research used a dataset of 25,051 slit lamp images from 3409 participants, paired with their corresponding physician-created medical reports. We used these data, split into training, validation, and test sets, to finetune the Bootstrapping Language-Image Pre-training framework toward report generation. The generated text reports and human-posed questions were then input into Llama2 for subsequent QA. We evaluated performance using qualitative metrics (including BLEU [bilingual evaluation understudy], CIDEr [consensus-based image description evaluation], ROUGE-L [Recall-Oriented Understudy for Gisting Evaluation—Longest Common Subsequence], SPICE [Semantic Propositional Image Caption Evaluation], accuracy, sensitivity, specificity, precision, and F1-score) and the subjective assessments of two experienced ophthalmologists on a 1-3 scale (1 referring to high quality). RESULTS We identified 50 conditions related to diseases or postoperative complications through keyword matching in initial reports. The refined slit lamp–GPT model demonstrated BLEU scores (1-4) of 0.67, 0.66, 0.65, and 0.65, respectively, with a CIDEr score of 3.24, a ROUGE (Recall-Oriented Understudy for Gisting Evaluation) score of 0.61, and a Semantic Propositional Image Caption Evaluation score of 0.37. The most frequently identified conditions were cataracts (22.95%), age-related cataracts (22.03%), and conjunctival concretion (13.13%). Disease classification metrics demonstrated an overall accuracy of 0.82 and an F1-score of 0.64, with high accuracies (≥0.9) observed for intraocular lens, conjunctivitis, and chronic conjunctivitis, and high F1-scores (≥0.9) observed for cataract and age-related cataract. For both report generation and QA components, the two evaluating ophthalmologists reached substantial agreement, with κ scores between 0.71 and 0.84. In assessing 100 generated reports, they awarded scores of 1.36 for both completeness and correctness; 64% (64/100) were considered “entirely good,” and 93% (93/100) were “acceptable.” In the evaluation of 300 generated answers to questions, the scores were 1.33 for completeness, 1.14 for correctness, and 1.15 for possible harm, with 66.3% (199/300) rated as “entirely good” and 91.3% (274/300) as “acceptable.” CONCLUSIONS This study introduces the slit lamp–GPT model for report generation and subsequent QA, highlighting the potential of large language models to assist ophthalmologists and patients.