Lung Segmentation using UNET in TensorFlow 2.0 | Chest X-Ray Segmentation
ฝัง
- เผยแพร่เมื่อ 17 ก.ย. 2024
- Hello friends. In this video, we are going to perform lung segmentation using the chest x-ray images dataset on the famous UNET architecture in TensorFlow 2.0 framework.
DATASET: academictorren...
CODE: github.com/nik...
UNET Implementation in TensorFlow: • UNET Architecture in T...
Blog Post:
- UNET Implementation in PyTorch: idiotdeveloper...
- RESUNET Implementation in PyTorch: idiotdeveloper...
- UNET Implementation in TensorFlow using Keras API: idiotdeveloper...
- What is UNET: idiotdeveloper...
- What is RESUNET: idiotdeveloper...
SUPPORT:
- / @idiotdeveloper
- www.buymeacoff...
FOLLOW ME:
BLOG: idiotdeveloper... sciencetonight...
FACEBOOK: / idiotdeveloper
TWITTER: / nikhilroxtomar
INSTAGRAM: instagram/nikhilroxtomar
PATREON: / idiotdeveloper
MY GEARS:
Intel i5-7400: amzn.to/3ilpq95
Gigabyte GA-B250M-D2V: amzn.to/3oPuntd
ZOTAC GeForce GTX 1060: amzn.to/2XNtsxn
LG 22MP68VQ 22 inch IPS Monitor: amzn.to/3soUKs5
Corsair VENGEANCE LPX 16GB: amzn.to/2LVyR2L
WD Green 240 GB SSD: amzn.to/3igt1Ft
1TB WD Blue: amzn.to/38I6uhw
Corsair VS550 550W: amzn.to/3nILHi3
Zebronics BT4440RUCF 4.1 Speakers: amzn.to/2XGu203
Segate 1TB Portable Hard Disk: amzn.to/3bF8YPG
Seagate Backup Plus Hub 8 TB External HDD: amzn.to/39wcqtj
Maono AU-A04 Condenser Microphone: amzn.to/35HHiWC
Techlicious 3.5mm Clip Microphone: amzn.to/3bERKSD
Redgear Dagger Headphones: amzn.to/3ssZNYr
hey bro,
your videos are really helpful to me. thank you, from Sri Lanka
Glad you like them!
Nice explanation, great job. Thank you
If possible please use white background, it will be easy for us to see character in white background...you r doing a great work..
Thanks for the valuable feedback 🙏
Thanks to you 🙏👍
Hi There! first of all I want to say that you're doing an amazing job! Also I want to ask that I'm looking for something that use image segmentation and subtract all other region of image leaving only left and right lung but with pattern inside lungs still prominent(not like in this video the output contains left and right lung but it is filled with white color not the patterns in the original image). and I want to apply it on a large dataset so that it produces my own customized dataset that I can use for further research!
Whoever replies, I want to say thank you that you stopped, considered my comment worth reading and helped me!
Again thanks all!
Hello u explained really well....can u please instruct on how you download this dataset...like how u used bit torrent client...I'm using windows 10...this can be really helpful for me
Can you explain how to convert the mask image into proper lung image again after segmentation?
Sir, any videos on computer aided diagnosis for lung cancer detection using LIDC IDRI dataset, which could help me out for my research.
Excellently explained. I tried the same code for my dataset. My images and labels are of size 512x512x1.
The batches are not properly created. The train_dataset shape is like this((None, 512, 512, 1), (None, 512, 512, 1)), types: (tf.float32, tf.float32)>
Can you pls help me in resolving this issue?
you can expand dims of your dataset to batch 1 by using np.expand_dim( data, axis= 0) this will convert your data to
( 512,512,1) -> (1,512,512,1)
Really u explain well.
Thanks for the valuable feedback.
thank you
Brother My images and masks is 2000x2000 please suggest how to train if possible How to resize them
Sir all I understand but one problem arise that is at last no picture is saving in result folder .result folder is empty.
If you are using Windows, please print the path where you are saving results. This problem mostly occurs in Windows
Thank you.Problem is solved.@@IdiotDeveloper
Sir in 11.30 ,is it ok to not to add lr in dice coefficient and iou func ?
Good job
hello, can u tell how u installed TensorFlow ...and also how u activated the environment?
Deployment of image segmentation via website plz make a video on this
how to implement dice loss on multi class classification
Sir I have some covid 19 positive cxr from kaggle but it doesn't contain mask images.. How will I extract mask image pliz help sir
You cannot extract the mask from images. An expert needs to annotate the image to get the mask.
@@IdiotDeveloper okay sir.. So sir can you recommend some segmentation architecture to segment lungs part from chest X ray which doesn't require mask image
Hai can u suggest code for super imposing one mask over another mask and make as one mask. If it possible please
from model import build_unet this is not correct module importing??? how to fix???
Check if there is any model.py file present.
Where do you use Data augmentation ?
Sir i have one question, I am trying to segment covid 19 Chest x ray but the dataset I downloaded from kaggle has no mask image but I want to segment the lungs. So sir can I use any dataset available on kaggle for lungs segmentation which contains mask images for training and then use my covid 19 dataset for testing data? Will it segment my covid dataset.. I really need your help sir.
If you only want to perform lung segmentation, then you can use any dataset containing proper images and masks, then test on your covid-19 x-ray images. But, if you want to do something like segmenting the covid -19 effected region in the lung x-ray images then this way would not work.
ho,can you tell me how we can done masking on chest x-ray?
Not only subscribe.
I share your videos too 🙂
Hi did you run the code?? I got error model.h5 not exist while loading model in eval.py
Hello sir....I need a dataset for lung nodule detection using CNN..could uu please help me to get the dataset
search in kaggle
my evaluated results are not saved .what can be the reason
The code is written on Ubuntu, maybe you are using Windows. So, just print all the path and make sure that are proper.
@@IdiotDeveloper yes l am on windows
I need covid-19 lung segmentation for my project with python code where can I get
hey, did you get the code, if yes , reply
@@profession4078 not yet still searching
@@swathiraju1552 try Unet segmentation , i am also looking for the code that integrates my dataset
@@profession4078 Hi. did you find it?
Hi. did you find it?
hi please make video on fcn architecture
Bro what is the need for this segmentation
Sir i having nlst dataset
I want to do heart segmentation for the data but i don't have train_y
For that i have done manual segmentation for 10 patients
And each patient is having 50 to 60 images
And after doing manual segmentation using
Robot.uk.vgg segmentation of Oxford
I have save the manual segmentation in .json fomat..
Can you tell know how to automatica segmentation of heart
I have even text you on instagram
Pls do share your linkedin profile too
Pls reply me
Sir can we use segmentation without GPU
Without GPU, training a deep learning model would be really difficult as it would take huge amount of time and heating issues would also be there.
What is the compiler ypu are using
I didn't get your question.
sir is it compulsory to have mask image for test data
For test data, mask is required to calculate the score (metrics). You can just use the image and predict the mask.
my man gonna do ISIC challenge again ?
Respected Sir,
From which side you download the dataset with their masks? I need brain tumour grayscale dataset with their masks(ground truth). Kindly help me for this because my whole project is pending for dataset. Plz help me.
I shall be very thankful to you for this kindness.
Regards.
For downloading various dataset, I use KAGGLE and academictorrents.com
@@IdiotDeveloper
Respected sir.
I have checked that site which you mention in the comment. But their no brain tumour dataset is available with ground truth. Kindly guide me that how I can download brain tumour dataset with their binary images.
@@IdiotDeveloper is it possible to do segmentation without mask file, means if i have only chest x-ray photos no mask of that data ... is there is any way do the segmentaion
If you don't have mask for the chest x-ray images, then it is not possible to segmentation.
@@IdiotDeveloper sir can we use mask image from some other lungs which is not for the image I will be using
Error : savemodel file does not exist at files/model.h5
The weigh file does not exist. You need to train the model.
@@IdiotDeveloper in training error value error :: test_size=0 should be either positive or smaller than a number of sample 0 or in float in the (0,1) range
Good video !
Can you do the implementation of InceptionResNetV2 please ?
You mean using InceptionResNetV2 as pretrained encoder in the UNET.
@@IdiotDeveloper Yes 🤩
I saw only on paper that there is a very good performance
do i need to get the mask when testing a new image ?
For predictions you only need image and if you want compare and calculate metrics you will need the ground truth mask also.
Can I detect lung cancer using this model?
I don't think so. This model is trained only for lung segmentation.