Hour 1: Planning for the Future (9:00 AM – 10:00 AM)

• Building a pipeline.

  • Importance of structuring a machine learning pipeline.

  • Understanding the flexibility and versatility of pipelines through interfacing and interchanging stages.

• Knowing your inputs and outputs.

  • Discussing the modality and dimensionality of input data.

  • Understanding how outputs vary depending on different tasks.

  • Exploring input and output sizes.

  • Use Case: IAN segmentation

Hour 2: Formulating the Problem (10:00 AM – 11:00 AM)

• Task definition

  • Defining the machine learning task, deep-learning.

  • Understanding losses and metrics.

• Preprocessing and Augmentations

  • Understanding transformations, their definitions, and how to visualize them.

  • Discussing the effects of 2D versus 3D dimensionality.

  • Exploring the attributes of different dimensions.

Hour 3: Training the Model (11:00 AM – 12 PM)

• What resources do I need?

  • Overview of necessary resources, including definitions, technologies, and types.

  • Introduction to DIY approaches for acquiring resources.

• What to do when I get stuck?

  • Strategies for hyperparameter tuning.

  • Understanding overfitting and using tools like Wandb for tracking experiments.

Hour 4: Hands-on Practice with Roboflow and ITK-Snap (12:00 PM - 1:00 PM)

• Introduction to Roboflow.

  • Steps for dataset preparation and augmentation.

  • How to create annotations specifically for dental images.

• Semiautomatic Segmentation

The objective of this workshop is to provide participants with a comprehensive understanding of the end-to-end process of building, training, and evaluating machine learning models, with a specific focus on dental imaging. The workshop will cover the essential stages, from planning and problem formulation to hands-on practice using industry-standard tools like Roboflow and ITK-Snap.

By the end of the workshop, participants will be equipped with practical skills to implement their own machine learning pipelines and troubleshoot common challenges.

Marwan Torki received the B.Sc. and M.Sc. degrees in computer science from Alexandria University, Egypt, in 2003 and 2006, respectively, and the Ph.D. degree in computer science from Rutgers University, NJ, USA, in 2011. He is an associate professor in the Computer and Systems Engineering Department at Alexandria University. His research interests include machine learning, computer vision, natural language processing, and deep learning. He published more than 90 publications and supervised more than 15 graduate students. He received many grants from the government and other industrial partners.

Marwa Baraka, lecturer of Pediatric Dentistry, Faculty of Dentistry, Alexandria University. She had her BS in 2010, MSc in 2015, and PhD in 2023 from the same university. She worked for six months as an adjunct assistant clinical professor in the Cariology, Restorative, and Endodontics department, School of Dentistry, University of Michigan, Ann Arbor, Michigan, USA. She is a Fulbright Alumni, cohort 2021, and had a non-degree PhD scholarship at the University of Michigan, Ann Arbor. She attained a US- Egypt cooperation grant from Venture Well and STDF and secured funding for her project: Pulp AI and was selected among five teams to have an incubation period in Georgia Tech, Atlanta, Georgia, USA to experience the US ecosystem. She has publications in top Q1 dental journals.

Mahmoud Gamal is a Teaching Assistant at the University of Alexandria, Egypt who received the B.Sc from the same insitute and currently finalising the M.Sc. degree. He also occupies various positions as a Senior Software Engineer and Researcher at KnightsLab

Participants are required to bring their personal laptops for this workshop.

The following are the recommended specifications for an optimal workflow during the workshop and for future practice and application:

Windows 10, Intel® Core™ i7 processor, RAM: 8 GB, solid state drive, 512GB.

• National dentists: 2000 EGP

• International dentists: 150 USD