Abstract:
Microscopy imaging aided with artificial intelligence (AI) and deep learning computer systems has brought about a revolution in how images can be consistently and efficiently recognised, verified and analysed. This project aimed to develop a web-based interface embedded with a supervised AI-aided deep learning computer system to enhance students’ learning of microscopy imaging knowledge and practical skills. We have developed a Convolutional Neural Network (CNN) model capable of identifying and classifying distinct organ tissue specimens with a testing accuracy of 96.64%. An automated cell counting platform was also developed for cell enumeration. A total of 148 participants were recruited to trial the learning platforms during five different workshops and practical lessons. Participant feedback shows that the majority expressed satisfaction with the lessons learned and their overall learning experience. Overall, our project had a positive impact on the participants’ perceived problem-solving and critical thinking skills through self-assessments of their own image specimens or a given image dataset. Participants also reported improvements in their microscopy image recognition skills and increased confidence in their practical skills after the lessons.
Code:
T0264
Principal Project Supervisors:
Keywords provided by authors:
- Microscopy imaging
- Cell density counting
- Artificial neural network (ANN)
- AI-generated feedback
- Cell biology laboratory skills
Start Date:
16 Jan 2023
End Date:
15 Jul 2024
Status:
Completed
Result:
This project successfully developed two web-based interfaces, i.e., AI-enabled platform for tissue recognition and automated cell counting platform with two L&T packages. A total of 148 participants were recruited for the learning trials and testing the platform in five different workshops/practical lessons. These include 97 EdUHK undergraduate students from two BEd(P) courses, INS3020 Living in the Information Age and HCS3064 Healthy Living; 15 undergraduate students from BEd(P)-GS and PGDE(P) programmes attended a General Studies Workshop on Microscopic World; 15 BEd(Science) students attended a workshop on “Microscope Image Recognition”; and 21 in-service teachers attended a PDP course, SCG5019 Effective Integration of IT for Scientific Inquiry and STEAM Education. We collected 122 participants’ feedback via questionnaire (response rate = 82.4% from 148 participants). The majority of participants expressed satisfaction with the lessons learned, with an average score of 3.54±0.56 (on a total score of 4.00). Our data also shows that BEd (Science) students who attended the lessons with the support of our AI-aided web-interface and self-assessments perceived that they improved their microscopy image recognition skills and gained more confident in organ tissue specimen differentiation after the lessons.
Impact:
This project aimed to address the challenges students face when learning microscopy skills. Often during image acquisition, students are unsure if they have captured the correct microscopy images and are unable to discern if anomalous images were obtained. Our project successfully developed an AI-supervised web-based interface that allows students to benefit from AI technology to verify and analyse their microscopy images with high accuracy. As shared by the students, they improved their microscopy practical skills and were satisfied with their overall learning experience when using this platform for self-assessment of their own microcopy images or given image dataset. Students provided positive feedback, noting that the online platform offered instant feedback after image uploads, enabled them to perform self-assessments via online quizzes, provided automated cell enumeration with self-correction functions, and facilitated more effective learning. Furthermore, feedback from in-service teachers indicates that our platform can also be used to teach secondary students biology and other related STEAM topics.
Financial Year:
2022-23
Type:
TDG