def main():
# Example use case of the synapse detection pipeline
# Location of Queries
queryFN = 'example_queries.json'
# List of Queries
list_of_queries = syn.loadQueriesJSON(queryFN)
# Test the first query
query = list_of_queries[0]
# Load the data
synaptic_volumes = da.load_tiff_from_query(query)
# Run Synapse Detection
# Takes ~5 minutes to run
result_vol = syn.getSynapseDetections(synaptic_volumes, query)
# Verify Output
label_vol = measure.label(result_vol > 0.9)
stats = measure.regionprops(label_vol)
print(len(stats)) # output should be 5440
def run_synapse_detection(atet_input):
"""
Run synapse detection and result evalution. The parameters need to be rethought
Parameters
-------------------
atet_input : dict
Returns
-------------------
output_dict : dict
"""
query = atet_input['query']
queryID = atet_input['queryID']
nQuery = atet_input['nQuery']
resolution = atet_input['resolution']
data_location = atet_input['data_location']
data_region_location = atet_input['data_region_location']
output_foldername = atet_input['output_foldername']
region_name = atet_input['region_name']
layer_mask_str = atet_input['mask_str']
dapi_mask_str = atet_input['dapi_mask_str']
mouse_number = atet_input['mouse_number']
# Load the data
synaptic_volumes = da.load_tiff_from_query(query, data_region_location)
# Load layer mask
if layer_mask_str != -1:
layer_mask = Image.open(layer_mask_str)
layer_mask = np.array(layer_mask)
# Load DAPI mask
dapi_mask_fn = os.path.join(dapi_mask_str,
str(mouse_number) + 'ss-DAPI-mask.tiff')
dapi_mask = da.imreadtiff(dapi_mask_fn)
# Merge DAPI mask and Layer 4 mask
if layer_mask_str != -1:
combined_mask = merge_DAPI_L4_masks(layer_mask, dapi_mask)
else:
dapi_mask = dapi_mask.astype(np.bool)
combined_mask = np.logical_not(dapi_mask) # keep portions without dapi
# Mask data
synaptic_volumes = mask_synaptic_volumes(synaptic_volumes, combined_mask)
volume_um3 = get_masked_volume(synaptic_volumes, combined_mask, resolution)
print(volume_um3)
# Run Synapse Detection
print('running synapse detection')
resultvol = syn.getSynapseDetections(synaptic_volumes, query)
# Save the probability map to file, if you want
outputNPYlocation = os.path.join(data_location, output_foldername,
region_name)
syn.saveresultvol(resultvol, outputNPYlocation, 'query_', queryID)
thresh = 0.9
queryresult = compute_measurements(resultvol, query, volume_um3, thresh)
output_dict = {
'queryID': queryID,
'query': query,
'queryresult': queryresult
}
return output_dict
def main():
"""
run site 3 synapse detection
"""
# Load metadata
metadata_fn = '/Users/anish/Documents/Connectome/SynapseAnalysis/data/M247514_Rorb_1/Site3Align2/site3_metadata.json'
#metadata_fn = 'site3_metadata_TdTomato.json'
metadata = syn.loadMetadata(metadata_fn)
datalocation = metadata['datalocation']
outputNPYlocation = metadata['outputNPYlocation']
query_fn = metadata['querylocation']
# List of Queries
listOfQueries = syn.loadQueriesJSON(query_fn)
print("Number of Queries: ", len(listOfQueries))
for n in range(0, len(listOfQueries)):
query = listOfQueries[n]
print(query)
# Load the data
synapticVolumes = da.loadTiffSeriesFromQuery(query, datalocation)
# Run Synapse Detection
# Takes ~5 minutes to run
resultVol = syn.getSynapseDetections(synapticVolumes, query)
# Save the probability map to file, if you want
syn.saveresultvol(resultVol, outputNPYlocation, 'resultVol', n)
# Save the thresholded results as annotation objects
# in a json file
# pd.probMapToJSON(resultVol, metadata, query, n)
"""
Evaluation of site3 synapse detection results
"""
# Load metadata
metadataFN = 'site3_metadata.json'
metadata = syn.loadMetadata(metadataFN)
outputJSONlocation = metadata['outputJSONlocation']
queryFN = metadata['querylocation']
evalparam = metadata['evalparam']
# List of Queries
listOfQueries = syn.loadQueriesJSON(queryFN)
listOfThresholds = []
listOfThresholds_to_text = []
listOfQueries_to_text = []
listofevals = []
thresh_list = [0.7, 0.8, 0.9]
# Evaluate each query individually
for n, query in enumerate(listOfQueries):
listOfThresholds.append(query['thresh'])
for thresh in thresh_list:
listOfThresholds_to_text.append(thresh)
query['thresh'] = thresh
print(query)
listOfQueries_to_text.append(query)
queryresult = pd.combineResultVolumes([n], [thresh], metadata,
evalparam)
listofevals.append(queryresult)
pd.printEvalToText(listofevals, listOfQueries_to_text,
listOfThresholds_to_text)
# Combine Queries
evaluation_parameters = metadata['evalparam']
pd.combineResultVolumes(list(range(0, len(listOfQueries))),
listOfThresholds, metadata, evaluation_parameters)