search

uci-cbcl / NoduleNet

[MICCAI' 19] NoduleNet: Decoupled False Positive Reduction for Pulmonary Nodule Detection and Segmentation
Other
185 stars 57 forks source link

The number of nodules #29

Open SherrySky97 opened 4 years ago

SherrySky97 commented 4 years ago

Hello. Thank you for your shared code. I have a problem about the number of nodules. I have used the LUNA16 data set, which contains 1186 nodules in 594 CT sets. Why you used 1,131 nodules from 583 CT sets？ Are there any differences in the selection criteria for the two data sets?

SherrySky97 commented 4 years ago

I ran your preprocessing program, but the images of mask.nrrd is all black, and the images of clean.nrrd is normal. The following is my display program：

path = "1.3.6.1.4.1.14519.5.2.1.6279.6001.100225287222365663678666836860_mask.nrrd" readdata, header = nrrd.read(path) print(readdata.shape) print(header) for i in range(0, readdata.shape[2]): nrrd_data, nrrd_options = nrrd.read(path) nrrd_image = Image.fromarray(nrrd_data[:,:,i]) nrrd_array = np.asarray(nrrd_image) nrrd_array = np.transpose(nrrd_array) cv2.imwrite( 'mask/'+ '%d' % i + ".jpg", nrrd_array)

I don't know why this problem occurs. Can you help me solve it？ Thanks.

SherrySky97 commented 4 years ago

I solved the second problem by adding nrrd_data = 255 * nrrd_data.astype('uint8')

SherrySky97 commented 4 years ago

I checked the contents of 1.3.6.1.4.1.14519.5.2.1.6279.6001.100225287222365663678666836860_bboxes.npy [[128.5 100.5 23.5 8. ] [200.5 63.5 255. 7. ]] However, in the annotations.csv of LUNA16, the diameters of the two nodules are 5.65 and 4.22 mm respectively. Do you know why the diameters are different？ And whether the different diameters affect the final result?

Jhd-git1 commented 2 years ago

I have a problem that Why the annotations.csv of LUNA16 is different from the 3_annotation,and how to visualize the detection of nodulenet?

cvbird commented 2 years ago

I checked the contents of 1.3.6.1.4.1.14519.5.2.1.6279.6001.100225287222365663678666836860_bboxes.npy [[128.5 100.5 23.5 8. ] [200.5 63.5 255. 7. ]] However, in the annotations.csv of LUNA16, the diameters of the two nodules are 5.65 and 4.22 mm respectively. Do you know why the diameters are different？ And whether the different diameters affect the final result?

Have you solved this issue yet ? Thank you.