Development and assessment of a novel multimedia-based educational software for teaching peripheral blood smear morphology | BMC Medical Education

Teaching software

The teaching software, a collaborative creation by PUMCH and XiaoYing, was incorporated into the multimedia teaching module as part of the HCI framework. This educational research necessitated only the use of a personal computer.

Design of the teaching software

The software’s instructional design encompasses three distinct modes: practice, competition, and test (Fig. 1). Concurrently, the software integrates a teaching gallery, facilitating users’ consolidation of their understanding of blood cell morphological characteristics by enabling a review of various blood cell images from their studies. Additionally, there are 30,000 annotated cells in the database available for teaching purposes. Each lesson (practice mode) includes 36 unique PBS single-cell images, organized into 18 pairs. Participants needed to accurately identify and select two identical cells from the non-repetitive cell map to eliminate the cells, while any incorrect pairing would leave the cells unchanged. The lesson chapter concludes successfully once all cells are accurately matched. A single practice session constitutes one lesson. For each correctly paired set of cells, the name of the cell type is displayed on the interface, thereby enhancing the user’s morphological comprehension of the blood cell type. At the end of the course, the software automatically compiles and analyzes the user’s learning outcomes. The practice mode is categorized into three tiers of difficulty: starter edition, advanced edition, and mastery edition, allowing users to choose their learning path based on their individual knowledge at baseline. The starter version encompasses seven clinically fundamental key cell types: band neutrophil, segmented neutrophil, eosinophil, basophil, monocyte, lymphocyte, and myeloblast (labeled as Blast No Lineage Spec in the software). The advanced version introduces an additional two cell types, specifically the lymphocyte variant form and metamyelocyte, building upon the starter version. Following this, the mastery version further incorporates three cell types: promyelocyte, myelocyte, and plasma cells, expanding upon the intermediate version’s content.

User interface. Homepage of CELLink

Upon the computation of their scores by the computer system, participants receive immediate, thorough feedback on their performance. Following the conclusion of a lesson or assessment, the teaching software undertakes an automatic evaluation and synthesis of responses. This process supports students in identifying, comprehending, and mastering cell types that are either challenging or perplexing, urging them to deliberate on and elucidate the characteristics of such cells. This deliberation is facilitated by examining the accuracy rates in identifying various PBSs and the misidentifications of cell types that were ambiguous, considering the chosen exercise’s level of difficulty. The software’s summary page records alterations in the users’ prior exercises, mirroring the students’ learning trajectories. Moreover, it compiles tailored revision suggestions for the cell types that were puzzling during the response process, thereby enabling students to rectify and bridge their knowledge deficiencies.

Design of the multimedia teaching lesson within the HCI framework

The CELLink teaching software encompasses a learning mode, a competition mode, and a test mode, as illustrated in Fig. 1. It features a teaching gallery (Fig. 2), which facilitates users in reinforcing their grasp of the morphological characteristics of blood cells through the review of various blood cell images pertinent to their studies. Figure 3 illustrates the learning interface of CELLink designed.

Learning interface by CEllink. There are 18 pairs of need to be matched. When one pair is matched correctly, the selected images will disappear. The course/lesson is over until all the images on the interface are eliminated

Upon completion of the practice mode, the software automatically processes and compiles the responses. This functionality aids students in identifying, understanding, and mastering cell types they find challenging or perplexing. It prompts students to engage in reflective thinking and to articulate the characteristics of such cells by evaluating their accuracy in recognizing different blood cell types and their mistakes in selecting ambiguous cell types, in the context of the exercise’s difficulty level. The summary page within the teaching software tracks modifications in users’ prior practices and mirrors the students’ learning progress (Appendix 1). Additionally, the software collates tailored review suggestions and furnishes students with a distinct compilation of incorrect pairings to scrutinize and bridge their knowledge gaps, thereby addressing the ambiguous cells encountered during the question-answering process (Fig. 4).

Wrong-matched image bank. The buttons at the top of the picture categorize the blood cells, allowing users to review misidentified cells. The number in the upper right corner of each individual cell picture represents the number of incorrect matches that occurred during the study

The competition mode within the CEllink teaching software utilizes a shared cell map among different users, enhancing interactivity and engagement (Fig. 5). The incorporation of a leaderboard, distinct from the practice mode, serves as an additional motivational tool by stimulating students’ interest in learning through fostering a sense of competition.

Summary of the competition: The ranking list summarizes the results of the current class’s PK competition

In the assessment mode, a distinct cell image database is employed to ensure that identical cells do not recur, thereby preventing users from “memorizing” information as a means to enhance their test performance.

Peripheral blood smear libraries and data collection

All blood cell images utilized in the teaching software are sourced from the PUMCH laboratory. The collection encompasses both typical and atypical peripheral blood cells to closely replicate the real clinical setting. Importantly, personal information such as patients’ names, genders, and diagnostic results are excluded from the dataset to ensure compliance with medical ethics. The data is exclusively used for the clinical laboratory education purposes of PUMCH.

The researchers photographed blood smears and assembled a database of PBSs, with the Sysmex SP-10 push-staining mechanism being utilized to prepare the PBSs. The preparation of blood smear materials, the smears themselves, and their staining were all conducted in accordance with production standards, as viewed under a low-power microscope. Furthermore, the “MarkServer” labeling system from XiaoYing Technology Co., Ltd. was employed for the manual screening and removal of blood cell images that were out of focus, exhibited abnormal staining, or lacked effective cell information in the images.

Data annotation

The cell identification and labeling under the microscope were conducted by an expert group consisting of two senior laboratory professionals from the Laboratory Department of PUMCH. The procedures for labeling and the objectivity of the labeling outcomes were rigorously controlled. In instances of discrepancy in cell labeling by the expert group, a process of cross-labeling was initiated. In case of differing opinions on labeling a specific cell, acknowledged PBS diagnosis experts within the field were consulted to review, mediate, and deliberate with the expert team to achieve a consensus on the result.

Study design

Twenty-six laboratory professionals, aged between 22 and 41 years, including 11 junior and 15 senior laboratory professionals, were enlisted for this study from the clinical laboratory of PUMCH. Personnel working in the field of peripheral blood smear morphology for less than five years are classified as junior professionals, while those with five or more years of experience are considered senior professionals.

The study design required participants to complete three tests: a pre-training test, a during training test, and a post-training test, with each test separated by two weeks. The pre-training test served as the baseline assessment. Following the initial study plan, the on-training test was scheduled after participants engaged with the teaching software for a prescribed 180 min bi-weekly for self-directed learning. This on-training assessment occurred two weeks after the commencement of the study. The final examination, the post-training test, was administered after continuing the established learning regimen, which again involved using the software for 180 min every two weeks, allowing for a consistent study pace before undertaking the third test.

Upon completing the three examinations, the instructional phase was concluded, during which each participant’s learning metrics—including usage data, accuracy rate, and temporal changes—were meticulously recorded and analyzed via computer. Subsequently, participants were asked to fill out an 11-item questionnaire aimed at gauging their user experience and evaluating any shifts in their self-efficacy. This questionnaire was designed to ascertain whether the teaching software adequately fulfilled the educational requirements it was intended to meet. Furthermore, it compiled insights on the software’s application in teaching contexts, impacts on learning interest, assessments of capability enhancement, interface design, and overall user experience. Through the deployment of anonymous surveys, the feedback and outcomes related to the educational content were gathered and scrutinized to assess the effectiveness of the teaching software’s utilization.

Statistical analysis

Prism 10 was employed for statistical analysis. The differences in responses to the “feedback questionnaire” before and after the training were analyzed using a paired t-test, comparing pre-training and post-training outcomes.