Reflekt

Reflekt aims to use simple camera will take photos of a user’s body every day, and algorithms will cull and crop the photos to create a profile for each mole. “So for each mole, there’ll be a historical record that shows what it looks like, the estimated size changes over time, discolorations, changes at the edges. Reflect could provide the doctor with a kind of dashboard that displays significant changes. “It might say that out of 15 moles on patient A’s body, two of them look like they have higher than that patient’s average mole growth in the last six months. This would give doctors more information and context than they get now with only annual or biennial visits.This would be like detecting early stage skin cancer.

Documentation

Project Description-Reflekt aims to use simple camera will take photos of a user’s body every day, and algorithms will cull and crop the photos to create a profile for each mole. “So for each mole, there will be a historical record that shows what it looks like, the estimated size changes over time, discolorations, changes at the edges. Reflect could provide the doctor with a kind of dashboard that displays significant changes. “It might say that out of 15 moles on patient A’s body, two of them look like they have higher than that patient’s average mole growth in the last six months. This would give doctors more information and context than they get now with only annual or biennial visits.This would be like detecting early stage skin cancer. With approximate 7% rise in skin cancer , this methodology would be a cost effective, time saving using a camera.

Methodology The setup would consists of a two-way mirror with a display monitor behind it, with the reflective ratio of the mirror allowing the user to see themselves as well as what’s on-screen behind the glass. The setup will also be equipped with a Raspberry Pi, a computer the size of a credit card on which pylon application will be built How to build pylon applications on Raspberry Pi

The idea is that Reflekt will take photos of a user’s body every day, and algorithms will cull and crop the photos to create a profile for each mole. “So for each mole, there will be a historical record that shows what it looks like, the estimated size changes over time, discolorations, changes at the edges Will have to simulate camera via machine learning to cover every fraction of human body via hands wide side ways and legs opened up to shoulder width pose.

With reference being screen behind mirror in which visible image (this will coincident with image formed by camera's on the screen)it would be accompanied by a green signal and buzzer(to acknowledge that successful image has been taken) for cameras to have every human body part (frontal , side , back)in focus.Once frontal part is fine , the sides and back will simply be turning around the same point.

Challenges- 1)For quick scan multiple cameras will be required and the data can be clubbed up for use. 2)Nude images need to be taken care off by coding protection. 3)Parts like inner thigh, arm pits imaging would be a challenge 4) 3D scan to detect moles frontal , side and back. One of the biggest challenges is to get enough of the right kind of data to train their models to identify moles accurately.

Code

GitHub Repository

https://github.com/egeozsoy/Mole_cancer_detection

Commits

egeozsoy
pushed 1ee9b357f47556fdf19f4c472631e68d9deb64c6
- change readme
2019-07-31 19:12:35 UTC
egeozsoy
pushed 2173c3a660a978cbb0f688c7de748385c6bbd9e6
- Further cleanup
2019-07-31 19:10:47 UTC
egeozsoy
pushed 12c7eed8701f1b25726bfb2ea91d75aef3b36850
- reformatting and cleanup
2019-07-31 19:05:07 UTC
egeozsoy
pushed 8762fd5a4975db6ba061221751631b1960433498
- 20 iterations
2019-07-30 19:42:48 UTC
egeozsoy
pushed 76f45b590a3b08ace2d7565bad332480e4904e7d
- catch edge case errors
2019-07-30 10:05:35 UTC
egeozsoy
pushed eefe543ee1ef67a334b286412b44366cff06867e
- export to core ml ready
2019-07-30 08:49:53 UTC
egeozsoy
pushed a857ce8714672fd746808684e680e867fcdcde3b
- activate bensprocessing
2019-07-29 22:22:13 UTC
egeozsoy
pushed 7245643af8efd757abaae81ce10385571446d9b4
- remove ban processing as it is not necessarily helping
2019-07-29 13:47:59 UTC
egeozsoy
pushed 85d36bed7bc24fea0d2b0a98718caf73b02f66e9
- add readme
2019-07-29 12:36:53 UTC
egeozsoy
pushed 5f586507f6aabb1a65e192ade1e3534f49c616f2
- add prediction part
2019-07-28 21:10:57 UTC
egeozsoy
pushed c332b4d917e1128e7f00f03c2bcba8d9d5af6feb
- save in folders
2019-07-28 15:31:15 UTC
egeozsoy
pushed b82c2f8d4bc9c1843b53e9d338a24c1df629bab7
- don't use batchsize 4
2019-07-28 09:34:08 UTC
egeozsoy
pushed 2882536520b18b631a859275673079cae561ac48
- better configs
2019-07-28 09:33:49 UTC
egeozsoy
pushed b50b0ad9861b5f458c4846058e392d282eb7f976
- implement small cnn for testing
2019-07-27 10:18:40 UTC
egeozsoy
pushed 81842c27d115e67c2f069167520b42f7fb30cdd9
- optional unfreeze paramater
2019-07-27 08:14:08 UTC
egeozsoy
pushed 3d76873ec4b292359c407cd0738267d020636d59
- training optional
2019-07-27 06:35:37 UTC
egeozsoy
pushed c4d7737f19bfad8a0eec41a8c5fd7a80a3db3e3b
- new comments
2019-07-27 05:48:15 UTC
egeozsoy
pushed ece999b96a1ecd6caa29eb172f9ace02a8982acb
- configurations file
2019-07-26 10:31:18 UTC
egeozsoy
pushed 0f8229db0c3935dc2fad34f02c4b4383b2674926
-clean new branch
2019-07-26 06:39:16 UTC
egeozsoy
pushed cb73b127f8a3e27e383e1b2d04692328314a96c8
-clean new branch
2019-07-26 06:34:05 UTC
egeozsoy
pushed 8867670b0e67bf4b234a697e16ca66fcbac1edd0
-clean new branch
2019-07-25 15:33:26 UTC
egeozsoy
pushed 96f5ea5f751bed65b0d63c154228b00d42c3d53b
- adjusted params so it overfits, can be controlled with dropout
2018-09-16 08:53:41 UTC
egeozsoy
pushed 6d6c79ac8d3272b5925f7af5dbde5ad11f7b094e
- this worked on cats vs dogs!
2018-09-01 11:55:58 UTC
egeozsoy
pushed 5a7da1416030a1fb59ac40b6c8656a949868e81b
- network changes
2018-08-28 05:56:52 UTC
egeozsoy
pushed 6f1f48749fc7e25b580751962f60f29d21c3737a
- network changes
2018-08-27 22:27:06 UTC
egeozsoy
pushed 11b249b7871ff8392e082ab4d8368f890ef990aa
- label image and retrain are tensorflow code
2018-08-27 12:48:31 UTC
egeozsoy
pushed 9264c574df0cba282987f183d236dc37e8dcd888
- resetted create_data
2018-08-26 19:03:24 UTC
egeozsoy
pushed 3480a87996a688beea64d81e65e6b603ef9f5fde
Initial commit
2018-08-25 19:34:25 UTC
Schematics
Title Description Format
Mole scan Way of scanning moles jpg
Bill of materials
Name Article number Link Quantity Unit Price
Rasberry Pi 1
UP embedded vision starter kit 1 349.00 $
Total 349.00 $
Info

Project State

Public Project

Licences

Software Licence: Project has no software
Hardware Licence: Project has no hardware

Admins

sharad-garg

Members

Does this project pique your interest?

Login or register to join or follow this project.

Comments
communitymanager report abuse
Hi! Reflekt sounds to get a good healthy vision support for doctors. What about other parts of body? Behind ears, belly, back, all sides of legs and so on? Did you see the other health-related projects here? Maybe you can come together and support each other. If you want, you can start a discussion in our forum.
sharad-garg report abuse
Hi , Thanks for the advice.Actually I have been sorting out the things..Every part needs to be focused , so the idea is to have an algorithm to analyse every part where in person is required to change positions like face recognition works...first recognition will be challenge. Also data protection is to be considered as naked images need to somehow coded .
communitymanager report abuse
You're welcome! Of course, data protection will be a huge topic, but maybe it's possible to integrate it into a kind of voluntarily self-service of a doctor's app or website. I guess people are more willing to use it if e.g. skin cancer is a topic in a family's history. Are you already working on the algorithm and recognition? What are you planning to do next?
sharad-garg report abuse
Hello. I am an instrumentation engineer who is bad at coding and looking for my Steve Wozniak...I know what and how to be done in terms of logics.
communitymanager report abuse
Use our forum and start a discussion. This will be the right place to ask for participants and coding support. Don't forget to link to your project.
au_chrismor report abuse
I have done some work in this space in the past, but the current range of cameras and SBC's make it much more practical. Count me in!
sharad-garg report abuse
Thanks. let’s reflekt
sharad-garg report abuse
Hi,please go through my idea and let me know if you can code it. I have to do it within 4 days for submission to contest second phase of smart life contest. I saw your profile and I am sure you can do it.
Back to top

Ready to join the project?

You'd like to participate ... Show more