Skip to content

Commit

Permalink
minor. revisions
Browse files Browse the repository at this point in the history
  • Loading branch information
@montcrypto
montcrypto committed Aug 20, 2023
1 parent 859f446 commit 247a113
Show file tree
Hide file tree
Showing 10 changed files with 552 additions and 318 deletions.
134 changes: 134 additions & 0 deletions .ipynb_checkpoints/001_model_builing-checkpoint.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,134 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "446c6e4f-a5f0-4e22-982b-1a90fa6c5742",
"metadata": {},
"source": [
"#### Codes for the papar by Tsuyama et al (2023) PlosOne\n",
"#### 001 : Data preparation for UNET"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "bf1fd544-7a20-46b5-a876-540f0a8f9387",
"metadata": {},
"outputs": [],
"source": [
"from common.data_prep import *\n",
"from common.model import *\n",
"# %matplotlib inline"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a7efcd6d-1bf3-4542-b08d-4801d999ef48",
"metadata": {},
"outputs": [],
"source": [
"# define your job number under _run directory\n",
"job_num = '003'\n",
"# set reduction_factor of the initial image \n",
"r_f = 2\n",
"# set unet training directory\n",
"train_set = ['image','mask']\n",
"#\n",
"set_dir(os.path.join('./_run',job_num))\n",
"set_dir(os.path.join('./_run',job_num,'model'))\n",
"for ids in train_set:\n",
" set_dir(os.path.join('./_run',job_num, 'train', ids))\n",
"set_dir(os.path.join('./_run',job_num,'segmentation')) # segmented VB areas in png\n",
"set_dir(os.path.join('./_run',job_num,'morphology')) # calculated VB morphological parameters\n",
"set_dir(os.path.join('./_run',job_num,'extracted_VB_images')) # extracted VBs along the radial positions"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "b36b0439-c00e-469c-be92-0dfc146e46ef",
"metadata": {},
"outputs": [],
"source": [
"# prepare training set for unet\n",
"path =\"./_original_png\"\n",
"filenumbers, Images, Labels = make_unet_data(path, r_f) \n",
"\n",
"n=0\n",
"for i in list(range(len(filenumbers))):\n",
" for k in list(range(filenumbers[i])):\n",
" n += 1\n",
" #cv2.cvtColor(img_bgrL, cv2.COLOR_BGR2GRAY)\n",
" cv２.imwrite(os.path.join('./_run',job_num, 'train', train_set[0],str(n)+'.png'), Images[i][k][:])\n",
" cv２.imwrite(os.path.join('./_run',job_num, 'train', train_set[1],str(n)+'.png'), Labels[i][k][:])\n",
"print(n, ' images were generated')"
]
},
{
"cell_type": "markdown",
"id": "5fbcf3de-093a-4811-947c-eea78a6e52cb",
"metadata": {},
"source": [
"The following code generates the model for segmentation of vasucular bundles "
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "44989d80-bdaf-4565-b31a-bce966f846d7",
"metadata": {},
"outputs": [],
"source": [
"data_gen_args = dict(rotation_range=1,\n",
" width_shift_range=0.1,\n",
" height_shift_range=0.1,\n",
" shear_range=0.1,\n",
" zoom_range=0.1,\n",
" horizontal_flip=True,\n",
" fill_mode='nearest')\n",
"myGene = trainGenerator(12,os.path.join('./_run',job_num, 'train'),'image','mask',data_gen_args,save_to_dir = None)\n",
"model = unet()\n",
"model_checkpoint = ModelCheckpoint(os.path.join('./_run',job_num, 'model','unet_moso_2.hdf5'), monitor='loss',verbose=1, save_best_only=True)\n",
"model.fit_generator(myGene,steps_per_epoch=300,epochs=20,callbacks=[model_checkpoint])"
]
},
{
"cell_type": "markdown",
"id": "0f150f33-2f42-46b1-a44b-fcb9238d44a0",
"metadata": {},
"source": [
"The end of codes"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8c48287a-b60f-4e95-a2b3-b9b93f0978d7",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "JS",
"language": "python",
"name": "js"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.7"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
107 changes: 107 additions & 0 deletions .ipynb_checkpoints/002_segment_analysis-checkpoint.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,107 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "3f934f2c-7259-4414-ac8f-6da7a42e76b7",
"metadata": {},
"source": [
"### 002: codes for segmentation"
]
},
{
"cell_type": "markdown",
"id": "734d6277-a4b9-44cf-af6e-0e19905c5dd4",
"metadata": {},
"source": [
"Read microscope images in the target directory, split them into patch images and predict VB area using the UNET model.\n",
"Generated segmented patches are then reassembled to the images and stored in r_target directory as original_filename+_predicted.png."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "c3babe8f-7b37-4024-a812-bcdac363c0bc",
"metadata": {},
"outputs": [],
"source": [
"from common.data_prep import *\n",
"from common.model import *\n",
"from common.img_patch import *\n",
"\n",
"\n",
"# target image for segmentation\n",
"target = './_test'\n",
"\n",
"# define your job number under _run directory\n",
"job_num = '001'\n",
"# set reduction_factor of the initial image \n",
"r_f = 2\n",
"\n",
"\n",
"set_dir(os.path.join('./_run',job_num,'segmentation')) # segmented VB areas of test images\n",
"#set_dir(os.path.join('./_run',job_num,'morphology')) # calculated VB morphological parameters\n",
"#set_dir(os.path.join('./_run',job_num,'extracted_VB_images')) # extracted VBs along the radial positions\n",
"\n",
" \n",
"target ='./_test'\n",
"r_target = os.path.join('./_run',job_num,'segmentation')\n",
"\n",
"\n",
"file_list=sorted([filename for filename in os.listdir(target) if not filename.startswith('.')])\n",
"\n",
"\n",
"for fln in range(len(file_list)):\n",
" img_path=os.path.join(target, file_list[fln])\n",
" res_path=os.path.join(r_target, file_list[fln])\n",
" img=io.imread(img_path, as_gray=True) # gray scale\n",
"\n",
" # r_f=2 #factor of image reduction\n",
" px=int(img.shape[0]/r_f)\n",
" py=int(img.shape[1]/r_f)\n",
" img_s=trans.resize(img, (px,py))\n",
" # making image patch of 512*512 )\n",
" testdata = patch_Gen(img_s)\n",
" pos = patch_Counter(img_s)\n",
" # predicton of vascular bundles by UNET \n",
" model = unet()\n",
" model.load_weights(os.path.join('./_run',job_num, 'model','unet_moso_2.hdf5'))\n",
" results = model.predict_generator(testdata,len(pos),verbose=1)\n",
" #saveResult(\"./prediction\",results)\n",
" # reassemble to the image\n",
" rec_img=patch_Assemble(img_s, results)\n",
" rec_img=img_as_ubyte(rec_img)\n",
" #　合成された画像を保存\n",
" cv2.imwrite(res_path.replace('.png', '_predicted.png'),rec_img)"
]
},
{
"cell_type": "markdown",
"id": "781e6731-93d7-4581-a6e6-07984aed3084",
"metadata": {},
"source": [
"The end of codes"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "JS",
"language": "python",
"name": "js"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.7"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Loading

0 comments on commit 247a113

Please sign in to comment.