def extract_candidates(predictions_scan, annotations, tf_matrix, pid, outputs_path):
print 'computing blobs'
start_time = time.time()
blobs = blobs_detection.blob_dog(predictions_scan[0, 0], min_sigma=1, max_sigma=15, threshold=0.1)
print 'blobs computation time:', (time.time() - start_time) / 60.
print 'n_blobs detected', len(blobs)
correct_blobs_idxs = []
for zyxd in annotations:
r = zyxd[-1] / 2.
distance2 = ((zyxd[0] - blobs[:, 0]) ** 2
+ (zyxd[1] - blobs[:, 1]) ** 2
+ (zyxd[2] - blobs[:, 2]) ** 2)
blob_idx = np.argmin(distance2)
print 'node', zyxd
print 'closest blob', blobs[blob_idx]
if distance2[blob_idx] <= r ** 2:
correct_blobs_idxs.append(blob_idx)
else:
print 'not detected !!!'
# we will save blobs the the voxel space of the original image
# blobs that are true detections will have blobs[-1] = 1 else 0
blobs_original_voxel_coords = []
for j in xrange(blobs.shape[0]):
blob_j = np.append(blobs[j, :3], [1])
blob_j_original = tf_matrix.dot(blob_j)
blob_j_original[-1] = 1 if j in correct_blobs_idxs else 0
if j in correct_blobs_idxs:
print 'blob in original', blob_j_original
blobs_original_voxel_coords.append(blob_j_original)
blobs = np.asarray(blobs_original_voxel_coords)
utils.save_pkl(blobs, outputs_path + '/%s.pkl' % pid)
def extract_candidates(predictions_scan, tf_matrix, pid, outputs_path):
print 'computing blobs'
start_time = time.time()
blobs = blobs_detection.blob_dog(predictions_scan[0, 0], min_sigma=1, max_sigma=15, threshold=0.1)
print 'blobs computation time:', (time.time() - start_time) / 60.
print 'n blobs detected:', blobs.shape[0]
blobs_original_voxel_coords = []
for j in xrange(blobs.shape[0]):
blob_j = np.append(blobs[j, :3], [1])
blob_j_original = tf_matrix.dot(blob_j)
blobs_original_voxel_coords.append(blob_j_original)
blobs = np.asarray(blobs_original_voxel_coords)
print blobs.shape
utils.save_pkl(blobs, outputs_path + '/%s.pkl' % pid)
def extract_candidates(predictions_scan, tf_matrix, pid, outputs_path):
print 'computing blobs'
start_time = time.time()
blobs = blobs_detection.blob_dog(predictions_scan[0, 0], min_sigma=1, max_sigma=15, threshold=0.1)
print 'blobs computation time:', (time.time() - start_time) / 60.
print 'n blobs detected:', blobs.shape[0]
blobs_original_voxel_coords = []
for j in xrange(blobs.shape[0]):
blob_j = np.append(blobs[j, :3], [1])
blob_j_original = tf_matrix.dot(blob_j)
blobs_original_voxel_coords.append(blob_j_original)
blobs = np.asarray(blobs_original_voxel_coords)
print blobs.shape
utils.save_pkl(blobs, outputs_path + '/%s.pkl' % pid)
def notateProcessor(self, notateQueue):
while True:
# get crop from queue
crop = notateQueue.get()
# blobs detection
image = exposure.equalize_hist(crop["image"])
print("notateProcessor", image.shape)
blobs = blobs_detection.blob_dog(image, threshold=0.3)
# get radius of blobs - now we have notation
notationList = np.zeros(blobs.shape)
for i in range(len(blobs)):
notationList[i] = self.calcRadius(blobs[i])
# get prob - now we have notation
notations = []
for i in range(len(notationList)):
probCube = crop["image"]
note = notationList[i]
z = int(note[0])
y = int(note[1])
x = int(note[2])
prob = probCube[z][y][x] / 2
r = note[3]
notation = np.array([z, y, x, r, prob])
notations.append(notation)
# get world coords
position = crop["position"]
metaFilename = crop["seriesuid"] + ".npy"
meta = self.serializer.readFromNpy("meta/", metaFilename)
worldCoord = meta["worldOrigin"]
steps = meta["steps"]
for i in range(len(notations)):
notations[i] = self.calcOffset(notations[i], position, steps)
notations[i] = self.calcWorldCoord(notations[i], worldCoord)
# write to disk
print("notateProcessor", crop["seriesuid"], crop["number"])
self.serializer.writeToNpy(
"notates/", "{0}-{1}.npy".format(crop["seriesuid"],
crop["number"]), notations)
def notate(self):
csvPath = self.dataPath + self.phaseSubPath + "csv/"
if not os.path.isdir(csvPath):
os.makedirs(csvPath)
with open(csvPath + "annotation.csv", "w") as file:
file.write("seriesuid,coordX,coordY,coordZ,diameter_mm\n")
with open(csvPath + "seriesuids.csv", "w") as file:
file.write("seriesuid\n")
pathList = glob(self.dataPath + self.phaseSubPath + "concat/*.npy")
for path in enumerate(tqdm(pathList)):
filename = os.path.basename(path[1])
seriesuid = filename.split(".")[0]
print(seriesuid)
data = self.serializer.readFromNpy("concat/", filename)
data = np.squeeze(data)
# notation filter with lung mask
mask = self.serializer.readFromNpy("mask/", filename)
data = data * mask
# blob dectection
data = exposure.equalize_hist(data)
blobs = blobs_detection.blob_dog(data,
threshold=0.3,
min_sigma=self.sqrt3)
# blobs = feature.blob_dog(data, threshold = 0.3)
# blobs = feature.blob_log(data)
# blobs = feature.blob_doh(data)
# get radius of blobs - now we have notation
notations = np.zeros(blobs.shape)
for i in range(len(blobs)):
notations[i] = self.calcRadius(blobs[i])
print(notations)
print(len(notations))
# convert to world coord
mhdFile = os.path.join(self.dataPath, self.phaseSubPath, "raw/",
seriesuid + ".mhd")
rawImage = sitk.ReadImage(mhdFile)
worldOrigin = np.array(rawImage.GetOrigin())[::-1]
for i in range(len(notations)):
notations[i] = self.calcWorldCoord(notations[i], worldOrigin)
# write notations to csv
csvPath = self.dataPath + self.phaseSubPath + "csv/"
if not os.path.isdir(csvPath):
os.makedirs(csvPath)
with open(csvPath + "annotation.csv", "w+") as file:
file.write("seriesuid,coordX,coordY,coordZ,diameter_mm\n")
for i in range(len(notations)):
line = "{0},{1},{2},{3},{4}\n".format(
seriesuid, notations[i][2], notations[i][1],
notations[i][0], notations[i][3] * 2.0)
file.write(line)
with open(csvPath + "seriesuids.csv", "w+") as file:
file.write("seriesuid")
for i in range(len(notations)):
file.write(seriesuid)