A diagnostic test of two-dimensional ultrasonic feature extraction based on artificial intelligence combined with blood flow Adler classification and contrast-enhanced ultrasound for predicting HER-2-positive breast cancer

Introduction

Breast cancer is one of the most common malignant tumors in women and has increased in incidence in recent years (1). Clinically, the treatment of breast cancer differs significantly depending of the molecular subtypes. Human epidermal growth factor receptor 2 (HER-2) is an important driver gene of breast cancer, and HER-2-positive breast cancer, due to its high malignancy, requires urgent early detection, diagnosis, and treatment (2,3). At present, the most common indirect prediction methods for the diagnosis of breast cancer are ultrasound, mammography and magnetic resonance imaging (MRI), however, it is still difficult to distinguish the different molecular types of breast cancer, including HER-2-positive breast cancer. Currently, with the continuous improvement of two-dimensional ultrasound, color Doppler ultrasound and contrast-enhanced ultrasound (CEUS), researchers have found that there is a certain correlation between ultrasonic imaging characteristics and molecular subtypes of breast cancer, including HER-2 positive breast cancer. However, the diagnostic accuracy of two-dimensional ultrasound, color Doppler ultrasound and CEUS for HER-2-positive breast cancer still needed to be improved, and the preoperative diagnosis of HER-2-positive breast cancer still need to rely on ultrasound-guided needle biopsy. Therefore, finding a model that can predict HER-2-positive breast cancer with high accuracy has become the focus of researchers. On the other hand, the application of ultrasound artificial intelligence (AI) analysis systems based on the Breast Imaging Reporting and Data System (BI-RADS) has become increasingly widespread, and the feature extraction of breast cancer two-dimensional ultrasound images has become similar to that of manual analysis. Especially in large-scale screening, AI analysis system would not reduce inspection efficiency and accuracy due to fatigue, and its diagnosis consistency is higher. Based on this, this study innovatively used AI analysis system to extract the features of two-dimensional ultrasound images of breast cancer in order to obtain more accurate experimental results (4-6). In addition, the Adler blood flow grade and CEUS features of breast nodules are often used for the classification in diagnosis (7-9). Based on this, this study aimed to establish a joint diagnosis prediction model for patients with HER-2-positive breast cancer by combining two-dimensional ultrasound feature extraction based on AI, blood flow Adler grade, and CEUS features in order to assist preoperative biopsy and provide a more precise preoperative diagnosis for patients with HER-2-positive breast cancer. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-2182/rc).

Methods

General information

A retrospective analysis was conducted on 140 patients with breast cancer who were hospitalized and underwent surgical treatment at the General Hospital of Xinjiang Military Command from January 2020 to November 2024. According to the postoperative pathological results, there were 88 patients with HER-2-positive breast cancer and 52 patients with HER-2-negative breast cancer. The patients' ages ranged from 29 to 63 years, with an average age of 45.96±8.73 years. These patients were divided into internal test samples and external validation samples in a ratio of 7:3 randomly. The internal test samples included 98 cases, which were 60 HER-2-positive cases and 38 HER-2-negative cases, the incidence of HER-2-positive was 61.22%. The external validation samples included 42 cases, which were 28 HER-2-positive cases and 14 HER-2-negative cases, the incidence of HER-2-positive was 66.67%. The internal test and external validation samples were divided into HER-2-positive group (60 cases in the internal test samples and 28 cases in the external validation samples) and HER-2-negative group (38 cases in the internal test samples and 14 cases in the external validation samples) according to the postoperative pathological results. The inclusion criteria were as follows: (I) complete clinical and surgical pathological data; (II) informed patient consent and approval by the hospital’s ethics committee; (III) completion of preoperative ultrasound AI analysis, color Doppler blood flow imaging, and CEUS examination; and (IV) all breast cancer nodules subjected to immunohistochemical examination. Meanwhile, the exclusion criteria were as follows: (I) pregnant or lactating patients, (II) nonnodular breast cancer, (III) nonsurgical treatment before surgery, and (IV) psychological or mental disorders leading to poor compliance. All ultrasound examinations were performed by ultrasound technicians with more than 5 years of experience who were proficient in all procedures involved in the study, and breast cancer molecular subtyping was based on the results of immunohistochemistry. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by ethics committee of General Hospital of Xinjiang Military Command (No. 20190036) and informed consent was taken from all the patients.

Two-dimensional ultrasound feature extraction of breast nodules with AI

Color Doppler ultrasound was conducted with a VINNO10 device (Vinno Technology, Suzhou, China) with a X6-16L linear probe (probe frequency of 6–16 MHz). The patient was placed on their back with their hands above their heads to fully expose the breast and axillary examination area. The AI real-time detection module built into the device was turned on, routine ultrasound examination of breast nodules was conducted, and the maximum cutface 2D image of the lesion was obtained. The AI real-time detection module wase used to intelligently delineate the boundary of the nodule and automatically detect the following 2D features: (I) longitudinal direction (parallel to the skin or not parallel to the skin), (II) shape (round, oval, or irregular), (III) margin (clear boundary or unclear boundary), (IV) posterior echo (enhanced, isoecho, or attenuated), (V) cystic/solid (cystic, solid, or cystic-solid mixed), and (VI) calcification (no calcification, microcalcification, or coarse calcification). Among those, not parallel to the skin, irregular shape, unclear boundary, posterior echo attenuated, solid or cystic-solid mixed, microcalcification or coarse calcification were treated as HER-2-positive, while others were treated as HER-2-negative. Finally, automated analysis was completed.

Blood flow Adler rating

The maximum 2D section image of breast nodule was obtained, and the device was adjusted to the color Doppler flow imaging mode for completion of color flow imaging of the target nodule. There are four levels of Adler grading for blood flow in nodules: level 0, no detectable or obvious blood flow in the nodules; level 1, only a few punctate blood flow signals detected in or around the nodules; level 2, a blood vessel with a length close to or beyond the radius of the nodule or a point-like blood flow signal at 3–4 places detectable in the nodule; level 3, two long blood vessels detectable in the nodule or two mesh blood flow signals detectable in or around the nodule. Among these, level 3 and level 4 were treated as HER-2-positive, while others were treated as HER-2-negative.

CEUS features

The maximum 2D section image of the breast nodule was obtained, the device was adjusted to the CEUS mode, and 4.8 mL if acoustic contrast agent of sulfur hexafluoride was injected into the patient through the median cubical vein to perform CEUS examination on the target nodule. Dynamic images were obtained to clarify the following CEUS characteristic information of the target nodule (compared with surrounding normal glandular tissue): (I) contrast enhancement level (low enhancement, equal enhancement, or high enhancement), (II) contrast agent enhancement speed (slow advance, equal advance, or fast forward), (III) contrast enhancement methods (centripetal, centrifugal, or diffuse), (IV) contrast agent distribution characteristics (uniform or uneven), (V) lesion range after CEUS (no increase or increase), (VI) puncture vessels (without or with perforating branches), and (VII) the nodule boundary after CEUS (clear or unclear). Among these, high enhancement, fast forward, centrifugal or diffuse, uneven, lesion range increased after CEUS, with perforating branches, unclear nodule boundary after CEUS were treated as HER-2-positive, while others were treated as HER-2-negative.

Statistical analysis

SPSS 22 statistical analysis software (IBM Corp., Armonk, NY, USA) was used for statistical analysis. The count data were expressed as the number of cases (n), the measurement data were expressed as mean ± standard deviation. Chi-square test was used to compare the count data between groups. Logistic regression analysis was used to establish a combined diagnosis model. The receiver operating characteristic (ROC) curves of the different diagnostic methods were plotted. The sensitivity, specificity, accuracy, kappa of different diagnostic methods were calculated and the area under the curve (AUC) values were calculated and compared. P<0.05 (two-sided) was considered statistically significant.

Results

Comparison of AI 2D ultrasonic feature information extraction and blood flow Adler grading in the two groups of internal test samples and external validation samples

In the 2D ultrasonic information extracted by AI, there was a statistically significant difference in the direction of nodule length diameter between the HER-2-positive group and the HER-2-negative group of internal test samples and external validation samples (P<0.05); specifically, the direction of nodule length diameter was not parallel to the skin in the HER-2-positive group, while the direction of nodule length diameter was parallel to the skin in the HER-2-negative group. Moreover, there was statistical significance in the Adler grade of intranodal blood flow between the positive and negative HER-2 groups of internal test samples and external validation samples (P<0.05). However, there was no significant difference between the positive and negative groups of internal test samples and external validation samples in nodule morphology, margin, posterior echo, cyst consolidation, calcification, or other characteristics (P>0.05; Table 1).

Comparison of CEUS characteristics between the two groups of internal test samples and external validation samples

In CEUS characteristic information, there were statistically significant differences in contrast agent distribution and nodule boundary after CEUS between the HER-2-positive group and HER-2-negative group of internal test samples and external validation samples (P<0.05). There was no statistical significance in contrast enhancement level, contrast enhancement speed, contrast enhancement method, lesion scope after CEUS, or puncture vessels in the lesion between the HER-2-positive group and the HER-2-negative group of internal test samples and external validation samples (P>0.05 and Table 2).

Establishment of the joint diagnosis model of internal test samples

The regression equation of the combined diagnosis model was constructed according to the extraction of 2D ultrasonic feature information (long diameter direction), blood flow Adler grading, and CEUS feature information (contrast agent distribution characteristics and nodule boundary after CEUS). The equation was as follows: Logit P = log [P/(1 – P)] = 1.414 × long diameter direction of nodules + 2.160 × blood flow Adler grade + 1.228 × contrast agent distribution characteristics + 1.742 × nodule boundary after CEUS – 3.115 (Table 3).

Diagnostic efficiency of the joint diagnostic model

The sensitivity, specificity, and accuracy of the combined diagnosis model, long diameter direction, blood flow Adler grading, contrast agent distribution characteristics, and nodule boundary after CEUS of internal test and external validation samples were respectively 78.95%, 85.00%, 82.62% and 92.86%, 96.34%, 95.24%; 68.42%, 75.00%, 72.45% and 64.28%, 82.14%, 76.19%; 63.16%, 71.67%, 68.37% and 71.43%, 89.29%, 83.33%; 63.16%, 70.00%, 67.34% and 64.29%, 85.71%, 78.57%; 68.42%, 68.33%, 68.37% and 71.43%, 85.71%, 80.95%; The features’ ranks in terms of kappa values were as follows: combined diagnosis model > contrast agent distribution characteristics > long diameter direction > nodule boundary after CEUS > blood flow Adler grading (Table 4). This indicates that combined diagnosis was more effective than was single-factor diagnosis in the overall diagnosis of breast cancer in the HER-2-positive group. The ROC curves of the different diagnostic methods were generated. The AUC value of the combined diagnosis model of internal test and external validation samples was 0.861 and 0.969, which was significantly higher (P<0.05) than that in the long diameter direction (0.717 and 0.732), blood flow Adler grade (0.674 and 0.786), CEUS distribution characteristics (0.666 and 0.750), and the nodule boundary after CEUS (0.684 and 0.786) by Delong test. There were statistically significant differences between the combined diagnosis model and the single-factor diagnosis method (P<0.05; Figure 1, Table 4).

Figure 1 ROC curve of combined diagnosis model, long diameter direction, blood flow Adler rating, CEUS distribution characteristics, and nodule boundary after CEUS of internal test samples and external validation samples. ROC, receiver operating characteristic; CEUS, contrast-enhanced ultrasound.

Discussion

The number of new cases of breast cancer worldwide in 2020 reached 2.26 million, with breast cancer surpassing lung cancer as the world’s most common cancer (10,11). Among these new cases, about 15% to 25% were HER-2-positive, and this type of breast cancer is highly invasive and has a high recurrence rate. Therefore, early detection, diagnosis, and treatment are advocated in clinical practice (12,13). As a convenient and rapid imaging examination method, ultrasound is one of the important modalities for the early screening of breast cancer. With the continuous promotion of ultrasonic AI technology, blood flow Adler grading, and CEUS technology, there has been a proliferation in the development of ultrasound diagnosis for breast cancer (14). However, it has been reported that the diagnostic classification of a single method is not satisfactory in identifying the genetic phenotype of breast cancer (15,16). Therefore, the early diagnosis of breast cancer with different gene phenotypes based on multimodal ultrasound has received increased attention in clinical research (17). Ultrasonic AI technology can quickly and accurately extract the 2D characteristics of breast cancer; blood flow Alder classification can classify blood flow characteristics of breast cancer; and CEUS can clarify the internal microvascular structure, blood supply status, and clearer boundary and morphological information by observing the diffusion of contrast agents in tumors (18-20). In this study, these commonly used ultrasound classification methods were selected to establish a more accurate HER-2-positive breast cancer detection model and provide a more accurate preoperative diagnostic basis for clinical practice (Figure 2).

Figure 2 Images of the independent risk factor and immunohistochemical results of HER-2-positive breast cancer. (A) Two-dimensional ultrasonic feature extraction based on AI (long diameter direction is not parallel to the skin), the yellow area was the two-dimensional ultrasonic feature extraction of breast nodule acquired by AI. (B) Blood flow Adler rating (level 2), the yellow area was the blood flow sampling area, and the red points were the blood flow signals. (C) CEUS features (the contrast agent is not evenly distributed, and the nodule boundary is not clear after angiography). (D) Immunohistochemical staining (20×). HER-2, human epidermal growth factor receptor 2; AI, artificial intelligence; CEUS, contrast-enhanced ultrasound.

In this study, long diameter direction and blood flow Adler grading were found to be the independent risk factors for HER-2-positive breast cancer, which may be due to the fact that breast cancer with this phenotype tends to have high proliferative activity and abundant blood supply and can spread and grow vertically beyond the normal tissue plane (21); this is consistent with the results of Hong et al. (22). In addition, contrast agent distribution characteristics and nodule boundaries after CEUS were also demonstrated to be the independent risk factors of HER-2-positive breast cancer, which may be explained by the ability of HER-2-positive breast cancer to grow rapidly, often necessitating the induction of additional neovascularization. The tumor parts with fewer new blood vessels would inducing ischemic and hypoxic necrosis due to the lack of nutrients required for tumor growth, which could cause the uneven distribution of contrast agent (23-28). Meanwhile, a tumor with greater neovascularization is highly invasive to the surrounding tissues, resulting in unclear nodule boundaries after CEUS (29-33). In this study, multifactor logistic regression was used to analyze the ultrasound imaging characteristics of HER-2-positive breast cancer, and a combined diagnosis model was established. The results showed that the diagnostic efficiency of the combined diagnosis model was significantly superior to that of single-factor diagnostic methods. This proposed method could provide a basis for the accurate preoperative diagnosis of HER-2-positive breast cancer.

Certain limitations to this study should be noted. Due to the small sample size of the study data, selection bias might have been present. In subsequent research, the sample size should be expanded, and other new ultrasound technologies, such as elastic imaging, should be integrated into the combined diagnostic model to further improve the prediction performance.

Conclusions

The combined AI-based model of 2D ultrasonic feature extraction, blood flow Adler grading, and CEUS for diagnosing HER-2–positive breast cancer can effectively improve diagnostic efficiency and provide a more accurate preoperative diagnosis for clinical treatment.

Acknowledgments

We would like to thank the director of the Department of Ultrasound, Tang Du Hospital, the Air Force Military Medical University, for providing the necessary support for this study. We would also like to thank the volunteers and patients for their cooperation throughout the entire study.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-2182/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-2182/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-2182/prf

Funding: This work was supported by General Program of National Natural Science Foundation of China (No. 82373343).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-2182/coif). All authors report funding from the General Program of National Nature Science Foundation of China (No. 82373343). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by ethics committee of General Hospital of Xinjiang Military Command (No. 20190036) and informed consent was taken from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. The Breast Cancer Professional Committee of the Chinese Anti-Cancer Association. Guidelines and Specifications for breast cancer Diagnosis and Treatment of China Anti-Cancer Association (2021 Edition). Chin Oncol 2021;31:954-1040.
2. Sitia L, Sevieri M, Signati L, et al. HER-2-Targeted Nanoparticles for Breast Cancer Diagnosis and Treatment. Cancers (Basel) 2022;14:2424. [Crossref] [PubMed]
3. Ferraro E, Drago JZ, Modi S. Implementing antibody drug conjugates (ADCs) in HER2-positive breast cancer: state of the art and future directions. Breast Cancer Res 2021;23:84. [Crossref] [PubMed]
4. Wang X, Li J, Zhao HD. Correlation between imaging features and molecular subtype in breast cancer. Journal of Chinese Practical Diagnosis and Therapy 2020;34:1071-3.
5. He X, Lu Y, Li J, et al. Development and validation of a prediction model for the diagnosis of breast cancer based on clinical and ultrasonic features. Gland Surg 2023;12:736-48. [Crossref] [PubMed]
6. Li JM, Shao YH, Sun XM, et al. Ultrasonic features of automated breast volume scanner (ABVS) and handheld ultrasound (HHUS) combined with molecular biomarkers in predicting axillary lymph node metastasis of clinical T1–T2 breast cancer. Quant Imaging Med Surg 2024;14:1359-68. [Crossref] [PubMed]
7. Hunter FW, Barker HR, Lipert B, et al. Mechanisms of resistance to trastuzumab emtansine (T-DM1) in HER-2 positive breast cancer. Br J Cancer 2020;122:603-12. [Crossref] [PubMed]
8. Liu Y, Wang Y, Wang Y, et al. Early prediction of treatment response to neoadjuvant chemotherapy based on longitudinal ultrasound images of HER2-positive breast cancer patients by Siamese multi-task network: A multicentre, retrospective cohort study. EClinicalMedicine 2022;52:101562. [Crossref] [PubMed]
9. Sigrist RMS, Liau J, Kaffas AE, et al. Ultrasound Elastography: Review of Techniques and Clinical Applications. Theranostics 2017;7:1303-29. [Crossref] [PubMed]
10. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
11. Kioutchoukova I, Lucke-Wold BP. Pyrotinib as a therapeutic for HER2-positive breast cancer. Transl Cancer Res 2023;12:1376-9. [Crossref] [PubMed]
12. Indini A, Rijavec E, Grossi F. Trastuzumab Deruxtecan: Changing the Destiny of HER2 Expressing Solid Tumors. Int J Mol Sci 2021;22:4774. [Crossref] [PubMed]
13. von Minckwitz G, Huang CS, Mano MS, et al. Trastuzumab Emtansine for Residual Invasive HER2-Positive Breast Cancer. N Engl J Med 2019;380:617-28. [Crossref] [PubMed]
14. Gu J, Zhong X, Fang C, et al. Deep Learning of Multimodal Ultrasound: Stratifying the Response to Neoadjuvant Chemotherapy in Breast Cancer Before Treatment. Oncologist 2024;29:e187-97. [Crossref] [PubMed]
15. Zheng X, Yao Z, Huang Y, et al. Deep learning radiomics can predict axillary lymph node status in early-stage breast cancer. Nat Commun 2020;11:1236. [Crossref] [PubMed]
16. Chen HP, Bap LY, Cheng YY, et al. Value of convergence sign on coronal plane images of automated breast volume scanning in diagnosis of breast lesions. Chinese Journal of Medical Ultrasound Celectronic Edition 2018;15:948-52.
17. Li FS, Yuan Q, Song CX, et al. Value of automatic breast volume scanning system in observing peripheral signs of breast tumors. Chinese Journal of Medical Ultrasound Celectronic Edition 2020;17:1183-8.
18. Zhang T, Tan T, Han L, et al. Predicting breast cancer types on and beyond molecular level in a multimodal fashion. NPJ Breast Cancer 2023;9:16. [Crossref] [PubMed]
19. Afshar P, Mohammadi A, Plataniotis KN, et al. From handcrafted to deep-learning-based cancer radiomics: challenges and opportunities. IEEE Signal Process Mag 2019;36:132-60. [Crossref]
20. Amioka A, Masumoto N, Gouda N, et al. Ability of contrast-enhanced ultrasonography to determine clinical responses of breast cancer to neoadjuvant chemotherapy. Jpn J Clin Oncol 2016;46:303-9. [Crossref] [PubMed]
21. Hurvitz SA, Martin M, Symmans WF, et al. Neoadjuvant trastuzumab, pertuzumab, and chemotherapy versus trastuzumab emtansine plus pertuzumab in patients with HER2-positive breast cancer (KRISTINE): a randomised, open-label, multicentre, phase 3 trial. Lancet Oncol 2018;19:115-26. [Crossref] [PubMed]
22. Hong AS, Rosen EL, Soo MS, et al. BI-RADS for sonography: positive and negative predictive values of sonographic features. AJR Am J Roentgenol 2005;184:1260-5. [Crossref] [PubMed]
23. Wang H, Zhan W, Chen W, et al. Sonography with vertical orientation feature predicts worse disease outcome in triple negative breast cancer. Breast 2020;49:33-40. [Crossref] [PubMed]
24. Liu SR, Liu C, Jing HM, et al. Subcapsular Injection of Ultrasonic Contrast Agent Distinguishes between Benign and Malignant Lymph Node Lesions Exhibiting Homogeneous Enhancement in Intravenous Contrast-Enhanced Ultrasound Images. Ultrasound Med Biol 2020;46:582-8. [Crossref] [PubMed]
25. Bocchi M, Sousa Pereira N, Furuya RK, et al. Expression of Ki67 and p53 Proteins: Breast Cancer Aggressivity Markers in Brazilian Young Patients. J Adolesc Young Adult Oncol 2021;10:379-88. [Crossref] [PubMed]
26. Cobleigh M, Yardley DA, Brufsky AM, et al. Baseline Characteristics, Treatment Patterns, and Outcomes in Patients with HER2-Positive Metastatic Breast Cancer by Hormone Receptor Status from SystHERs. Clin Cancer Res 2020;26:1105-13. [Crossref] [PubMed]
27. Wu VS, Kanaya N, Lo C, et al. From bench to bedside: What do we know about hormone receptor-positive and human epidermal growth factor receptor 2-positive breast cancer? J Steroid Biochem Mol Biol 2015;153:45-53. [Crossref] [PubMed]
28. Gradishar WJ, Moran MS, Abraham J, et al. NCCN Guidelines® Insights: Breast Cancer, Version 4.2023. J Natl Compr Canc Netw 2023;21:594-608. [Crossref] [PubMed]
29. He Y, Wang MF, Zheng SS, et al. Correlation analysis of microcalcification in breast cancer with HER 2, ER, PR, and Ki67(J). Chinese Journal of Ultrasound in Medicine 2022;38:633-6.
30. Ding T, Kong XH, Yang YQ. Ultrasonographic features of four molecular subtypes breast cancer. Chin J Mod Med 2021;31:24-8.
31. Sha R, Dong X, Yan S, et al. Cuproptosis-related genes predict prognosis and trastuzumab therapeutic response in HER2-positive breast cancer. Sci Rep 2024;14:2908. [Crossref] [PubMed]
32. Tang M, Rong Y, Li X, et al. Anoikis-related genes in breast cancer patients: reliable biomarker of prognosis. BMC Cancer 2024;24:1163. [Crossref] [PubMed]
33. Mendiburu-Eliçabe M, García-Sancha N, Corchado-Cobos R, et al. NCAPH drives breast cancer progression and identifies a gene signature that predicts luminal a tumour recurrence. Clin Transl Med 2024;14:e1554. [Crossref] [PubMed]