init

Q1/code.py

@@ -0,0 +1,97 @@
+# code.py
+# This is an image preprocessing code for HandBoneXRay data, please adopt env.yaml to
+# create conda env to run this script
+#
+# author  : deng
+# date    : 20230919
+# platform: MacBook Pro 14 2021
+
+import os
+from shutil import rmtree
+
+import cv2
+import pydicom
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+if __name__ == '__main__':
+
+    xray_dir = 'HandBoneXRay/'
+    result_dir = 'results'
+    plot = False
+
+    # Create result dir to save processed images
+    if os.path.isdir(result_dir):
+        rmtree(result_dir)
+    os.makedirs(result_dir)
+
+    # Get xray dicom paths
+    dicom_paths = [os.path.join(xray_dir, file_name)
+                  for file_name in os.listdir(xray_dir)
+                  if file_name.endswith('dcm')]
+
+    for dicom_path in dicom_paths:
+        print(f'Start to process {dicom_path}')
+
+        # Read image
+        ds = pydicom.dcmread(dicom_path)
+        pixel_array = ds.pixel_array
+        orig_pixel_array = pixel_array.copy()
+        if plot:
+            plt.title('Raw image')
+            plt.imshow(pixel_array, cmap='gray')
+            plt.show()
+
+        # Thresholding
+        pixel_array[pixel_array < 1] = 0
+        pixel_array[pixel_array >= 1] = 1
+        pixel_array = pixel_array.astype(np.uint8)
+        if plot:
+            plt.title('Thresholding')
+            plt.imshow(pixel_array, cmap='gray')
+            plt.show()
+
+        # Calculate rotation angle by minAreaRect method
+        contours, _ = cv2.findContours(pixel_array, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
+        contour_w_max_area = max(contours, key = cv2.contourArea)
+        rect = cv2.minAreaRect(contour_w_max_area)
+        angle = rect[2]
+        if angle > 45:
+            angle -= 90
+
+        # Rotate image
+        h, w = orig_pixel_array.shape[:2]
+        center = rect[0]
+        matrix = cv2.getRotationMatrix2D(center, angle, 1.0)
+        rotated = cv2.warpAffine(orig_pixel_array, matrix, (w, h), flags=cv2.INTER_CUBIC, borderMode=cv2.BORDER_CONSTANT)
+        if plot:
+            plt.title('Rotated image')
+            plt.imshow(rotated, cmap='gray')
+            plt.show()
+
+        # Crop edge
+        xs, ys = np.nonzero(rotated)
+        x_min,x_max = xs.min(), xs.max()
+        y_min, y_max = ys.min(), ys.max()
+        cropped = rotated[x_min:x_max+1, y_min:y_max+1]
+        cropped = cropped.astype(np.uint16)
+        if plot:
+            plt.title('Cropped image')
+            plt.imshow(cropped, cmap='gray')
+            plt.show()
+
+        # Save it
+        save_path = os.path.join(result_dir, os.path.basename(dicom_path))
+        ds.PixelData = cropped
+        ds.Rows = cropped.shape[0]
+        ds.Columns = cropped.shape[1]
+        pydicom.filewriter.dcmwrite(save_path, ds)
+        if plot:
+            ds = pydicom.dcmread(save_path)
+            pixel_array = ds.pixel_array
+            plt.title('Saved image')
+            plt.imshow(pixel_array, cmap='gray')
+            plt.show()
+
+        print('done.')

Q1/env.yaml

@@ -0,0 +1,134 @@
+name: Q1
+channels:
+  - conda-forge
+dependencies:
+  - aom=3.5.0
+  - brotli=1.1.0
+  - brotli-bin=1.1.0
+  - bzip2=1.0.8
+  - c-ares=1.19.1
+  - ca-certificates=2023.7.22
+  - cairo=1.16.0
+  - certifi=2023.7.22
+  - contourpy=1.1.1
+  - cycler=0.11.0
+  - dav1d=1.2.1
+  - expat=2.5.0
+  - ffmpeg=6.0.0
+  - font-ttf-dejavu-sans-mono=2.37
+  - font-ttf-inconsolata=3.000
+  - font-ttf-source-code-pro=2.038
+  - font-ttf-ubuntu=0.83
+  - fontconfig=2.14.2
+  - fonts-conda-ecosystem=1
+  - fonts-conda-forge=1
+  - fonttools=4.42.1
+  - freetype=2.12.1
+  - fribidi=1.0.10
+  - gettext=0.21.1
+  - gmp=6.2.1
+  - gnutls=3.7.8
+  - graphite2=1.3.13
+  - harfbuzz=8.2.0
+  - hdf5=1.14.2
+  - icu=73.2
+  - jasper=4.0.0
+  - kiwisolver=1.4.5
+  - krb5=1.21.2
+  - lame=3.100
+  - lcms2=2.15
+  - lerc=4.0.0
+  - libabseil=20230802.1
+  - libaec=1.0.6
+  - libass=0.17.1
+  - libblas=3.9.0
+  - libbrotlicommon=1.1.0
+  - libbrotlidec=1.1.0
+  - libbrotlienc=1.1.0
+  - libcblas=3.9.0
+  - libcurl=8.3.0
+  - libcxx=16.0.6
+  - libdeflate=1.19
+  - libedit=3.1.20191231
+  - libev=4.33
+  - libexpat=2.5.0
+  - libffi=3.4.2
+  - libgfortran=5.0.0
+  - libgfortran5=13.2.0
+  - libglib=2.78.0
+  - libiconv=1.17
+  - libidn2=2.3.4
+  - libjpeg-turbo=2.1.5.1
+  - liblapack=3.9.0
+  - liblapacke=3.9.0
+  - libnghttp2=1.52.0
+  - libopenblas=0.3.24
+  - libopencv=4.8.0
+  - libopenvino=2023.0.2
+  - libopenvino-arm-cpu-plugin=2023.0.2
+  - libopenvino-auto-batch-plugin=2023.0.2
+  - libopenvino-auto-plugin=2023.0.2
+  - libopenvino-hetero-plugin=2023.0.2
+  - libopenvino-ir-frontend=2023.0.2
+  - libopenvino-onnx-frontend=2023.0.2
+  - libopenvino-paddle-frontend=2023.0.2
+  - libopenvino-pytorch-frontend=2023.0.2
+  - libopenvino-tensorflow-frontend=2023.0.2
+  - libopenvino-tensorflow-lite-frontend=2023.0.2
+  - libopus=1.3.1
+  - libpng=1.6.39
+  - libprotobuf=4.23.4
+  - libsqlite=3.43.0
+  - libssh2=1.11.0
+  - libtasn1=4.19.0
+  - libtiff=4.6.0
+  - libunistring=0.9.10
+  - libvpx=1.13.0
+  - libwebp-base=1.3.2
+  - libxcb=1.15
+  - libxml2=2.11.5
+  - libzlib=1.2.13
+  - llvm-openmp=16.0.6
+  - matplotlib=3.8.0
+  - matplotlib-base=3.8.0
+  - munkres=1.1.4
+  - ncurses=6.4
+  - nettle=3.8.1
+  - numpy=1.26.0
+  - opencv=4.8.0
+  - openh264=2.3.1
+  - openjpeg=2.5.0
+  - openssl=3.1.2
+  - p11-kit=0.24.1
+  - packaging=23.1
+  - pcre2=10.40
+  - pillow=10.0.1
+  - pip=23.2.1
+  - pixman=0.40.0
+  - pthread-stubs=0.4
+  - pugixml=1.13
+  - py-opencv=4.8.0
+  - pydicom=2.4.3
+  - pyparsing=3.1.1
+  - python=3.10.12
+  - python-dateutil=2.8.2
+  - python_abi=3.10
+  - readline=8.2
+  - setuptools=68.2.2
+  - six=1.16.0
+  - snappy=1.1.10
+  - svt-av1=1.7.0
+  - tbb=2021.10.0
+  - tk=8.6.12
+  - tornado=6.3.3
+  - tzdata=2023c
+  - unicodedata2=15.0.0
+  - wheel=0.41.2
+  - x264=1!164.3095
+  - x265=3.5
+  - xorg-libxau=1.0.11
+  - xorg-libxdmcp=1.1.3
+  - xz=5.2.6
+  - zlib=1.2.13
+  - zstd=1.5.5
+prefix: /Users/xiao_deng/miniforge3/envs/Q1

Q2/code.py

@@ -0,0 +1,63 @@
+# code.py
+# Please use env.yaml to build corresponding conda env if you want to run this script
+#
+# author  : deng
+# date    : 20230919
+# platform: MacBook Pro 14 2021
+
+VALID_CHARS = ['T', 'F', 'O', 'A', '(', ')']
+CAL_DICT = {
+    'T': 'T',
+    'F': 'F',
+    'TAT': 'T',
+    'TAF': 'F',
+    'FAT': 'F',
+    'FAF': 'F',
+    'TOT': 'T',
+    'TOF': 'T',
+    'FOT': 'T',
+    'FOF': 'F'
+}
+
+
+def log_cal(exp: str) -> str:
+    while True:
+        if exp in CAL_DICT:
+            return CAL_DICT[exp]
+        else:
+            try:
+                exp = CAL_DICT[exp[:3]] + exp[3:]
+            except KeyError:
+                return 'E'
+
+
+def bool_exp(exp: str) -> str:
+    stack = []
+
+    for char in exp:
+        if char not in VALID_CHARS:
+            return 'E'
+        if char == ')':
+            sub_exp = ''
+            while True:
+                try:
+                    val = stack.pop()
+                    if val == '(':
+                        break
+                    sub_exp = val + sub_exp
+                except IndexError:
+                    return 'E'
+            cal_exp = log_cal(sub_exp)
+            if cal_exp == 'E':
+                return 'E'
+            stack.append(cal_exp)
+        else:
+            stack.append(char)
+
+    return log_cal(''.join(stack))
+
+
+if __name__ == '__main__':
+
+    exp = input('Please input expression:').upper()
+    print(bool_exp(exp))

Q2/env.yaml

@@ -0,0 +1,20 @@
+name: Q2
+channels:
+  - conda-forge
+dependencies:
+  - bzip2=1.0.8
+  - ca-certificates=2023.7.22
+  - libffi=3.4.2
+  - libsqlite=3.43.0
+  - libzlib=1.2.13
+  - ncurses=6.4
+  - openssl=3.1.2
+  - pip=23.2.1
+  - python=3.10.12
+  - readline=8.2
+  - setuptools=68.2.2
+  - tk=8.6.12
+  - tzdata=2023c
+  - wheel=0.41.2
+  - xz=5.2.6
+prefix: /Users/xiao_deng/miniforge3/envs/Q2

Q3/code.py

@@ -0,0 +1,52 @@
+# code.py
+# Please use env.yaml to build corresponding conda env if you want to run this script
+#
+# author  : deng
+# date    : 20230919
+# platform: MacBook Pro 14 2021
+
+import pandas
+from sklearn import metrics
+import matplotlib.pyplot as plt
+
+
+if __name__ == '__main__':
+
+    csv_path = 'psudo_result.csv'
+    df = pandas.read_csv(csv_path)
+    df = df[:-1]
+
+    # Quick glance
+    print('Ground truth')
+    print(df['Ground truth'].value_counts())
+    print('Gender')
+    print(df['Gender'].value_counts())
+    print('Age')
+    plt.title('All patients')
+    plt.xlabel('age')
+    plt.ylabel('count')
+    df['Age'].hist(bins=20)
+    plt.show()
+    plt.title('Sick patients')
+    plt.xlabel('age')
+    plt.ylabel('count')
+    df[df['Ground truth'] == 'Sick']['Age'].hist(bins=20)
+    plt.show()
+
+    # Confusion Matrix
+    # threshold = 0.5
+    # gt = df['Ground truth']
+    # pred = ['Sick' if val >= threshold else 'No Sick'
+    #         for val in df['AI pred']]
+    # print(metrics.classification_report(gt, pred))
+
+    # ROC curve
+    gt = [1 if val == 'Sick' else 0
+          for val in df['Ground truth']]
+    pred = df['AI pred']
+    fpr, tpr, thresholds = metrics.roc_curve(gt, pred, pos_label=1)
+    plt.title('ROC curve')
+    plt.plot(fpr, tpr)
+    plt.ylabel('True Positive Rate')
+    plt.xlabel('False Positive Rate')
+    plt.show()

Q3/env.yaml

@@ -0,0 +1,71 @@
+name: Q3
+channels:
+  - conda-forge
+dependencies:
+  - brotli=1.1.0
+  - brotli-bin=1.1.0
+  - bzip2=1.0.8
+  - ca-certificates=2023.7.22
+  - certifi=2023.7.22
+  - contourpy=1.1.1
+  - cycler=0.11.0
+  - fonttools=4.42.1
+  - freetype=2.12.1
+  - joblib=1.3.2
+  - kiwisolver=1.4.5
+  - lcms2=2.15
+  - lerc=4.0.0
+  - libblas=3.9.0
+  - libbrotlicommon=1.1.0
+  - libbrotlidec=1.1.0
+  - libbrotlienc=1.1.0
+  - libcblas=3.9.0
+  - libcxx=16.0.6
+  - libdeflate=1.19
+  - libffi=3.4.2
+  - libgfortran=5.0.0
+  - libgfortran5=13.2.0
+  - libjpeg-turbo=2.1.5.1
+  - liblapack=3.9.0
+  - libopenblas=0.3.24
+  - libpng=1.6.39
+  - libsqlite=3.43.0
+  - libtiff=4.6.0
+  - libwebp-base=1.3.2
+  - libxcb=1.15
+  - libzlib=1.2.13
+  - llvm-openmp=16.0.6
+  - matplotlib=3.8.0
+  - matplotlib-base=3.8.0
+  - munkres=1.1.4
+  - ncurses=6.4
+  - numpy=1.26.0
+  - openjpeg=2.5.0
+  - openssl=3.1.2
+  - packaging=23.1
+  - pandas=2.1.0
+  - pillow=10.0.1
+  - pip=23.2.1
+  - pthread-stubs=0.4
+  - pyparsing=3.1.1
+  - python=3.10.12
+  - python-dateutil=2.8.2
+  - python-tzdata=2023.3
+  - python_abi=3.10
+  - pytz=2023.3.post1
+  - readline=8.2
+  - scikit-learn=1.3.0
+  - scipy=1.11.2
+  - setuptools=68.2.2
+  - six=1.16.0
+  - threadpoolctl=3.2.0
+  - tk=8.6.12
+  - tornado=6.3.3
+  - tzdata=2023c
+  - unicodedata2=15.0.0
+  - wheel=0.41.2
+  - xorg-libxau=1.0.11
+  - xorg-libxdmcp=1.1.3
+  - xz=5.2.6
+  - zstd=1.5.5
+prefix: /Users/xiao_deng/miniforge3/envs/Q3

Q3/psudo_result.csv

@@ -0,0 +1,102 @@
+Sample ID,Gender,Age,AI pred,Ground truth
+S_0,Male,42.0,0.183843222651211,No Sick
+S_1,Male,54.0,0.032451971485353635,No Sick
+S_2,Male,27.0,0.1649091208860276,No Sick
+S_3,Female,51.0,0.9838273079563034,Sick
+S_4,Female,32.0,0.7008005948749091,No Sick
+S_5,Female,50.0,0.04646539805577132,No Sick
+S_6,Female,63.0,0.37605963432485756,No Sick
+S_7,Male,48.0,0.5221502126827561,Sick
+S_8,Female,43.0,0.6380703073440606,No Sick
+S_9,Male,50.0,0.16000187268400592,No Sick
+S_10,Female,33.0,0.8684964025015954,No Sick
+S_11,Female,52.0,0.6292457724763052,No Sick
+S_12,Male,49.0,0.4549646243576516,Sick
+S_13,Female,26.0,0.5097453569913539,No Sick
+S_14,Female,60.0,0.3337711914417254,No Sick
+S_15,Female,48.0,0.019190399941885428,No Sick
+S_16,Male,47.0,0.04086642173184299,No Sick
+S_17,Female,54.0,0.18241359146528532,No Sick
+S_18,Male,58.0,0.32089169346089924,No Sick
+S_19,Female,36.0,0.413237627716878,Sick
+S_20,Female,48.0,0.13062595790437023,No Sick
+S_21,Female,56.0,0.29018011048552517,No Sick
+S_22,Female,53.0,0.2555372095893874,No Sick
+S_23,Female,44.0,0.09580993903210833,No Sick
+S_24,Male,55.0,0.6877916599706746,Sick
+S_25,Male,56.0,0.40153291103181854,Sick
+S_26,Male,44.0,0.17088154472476824,No Sick
+S_27,Male,60.0,0.28903419437712474,No Sick
+S_28,Male,39.0,0.3059480568045402,No Sick
+S_29,Female,51.0,0.30293016203624257,No Sick
+S_30,Male,32.0,0.5986074572752382,Sick
+S_31,Male,61.0,0.1679725784069874,No Sick
+S_32,Female,20.0,0.19933363134841395,No Sick
+S_33,Female,75.0,0.2936272223993208,No Sick
+S_34,Female,60.0,0.38458266378064254,No Sick
+S_35,Female,36.0,0.5476216653350379,No Sick
+S_36,Female,22.0,0.2540285986407038,No Sick
+S_37,Female,41.0,0.30688487550042554,No Sick
+S_38,Male,58.0,0.010881451914966034,No Sick
+S_39,Female,57.0,0.3141640693118162,No Sick
+S_40,Female,57.0,0.19447009952931774,No Sick
+S_41,Male,49.0,0.33065208227575404,Sick
+S_42,Female,59.0,0.751020059639999,Sick
+S_43,Male,73.0,0.2874468176061555,No Sick
+S_44,Male,47.0,0.12089859681706382,No Sick
+S_45,Male,66.0,0.17528475949731379,No Sick
+S_46,Female,64.0,0.8819982427894725,No Sick
+S_47,Female,42.0,0.5005003573463266,No Sick
+S_48,Female,43.0,0.2745791003849961,No Sick
+S_49,Female,45.0,0.4161238121442544,No Sick
+S_50,Male,35.0,0.41528904708151504,No Sick
+S_51,Female,52.0,0.8146647348503631,No Sick
+S_52,Male,49.0,0.17604193457147554,No Sick
+S_53,Male,53.0,0.05322796458244725,No Sick
+S_54,Female,45.0,0.4299485265172557,No Sick
+S_55,Female,63.0,0.35375757738991287,No Sick
+S_56,Male,37.0,0.2164630863950512,No Sick
+S_57,Female,43.0,0.7542813473217973,No Sick
+S_58,Male,40.0,0.12277033509439539,No Sick
+S_59,Male,57.0,0.028993766542943746,No Sick
+S_60,Male,63.0,0.1840667841524664,No Sick
+S_61,Female,42.0,0.9182713866681889,No Sick
+S_62,Female,38.0,0.4997258327162861,No Sick
+S_63,Female,27.0,0.057822477359941504,No Sick
+S_64,Male,45.0,0.2869275808238832,No Sick
+S_65,Male,56.0,0.27075663892334423,No Sick
+S_66,Female,78.0,0.5147121232117046,Sick
+S_67,Male,57.0,0.4303055422658295,Sick
+S_68,Female,65.0,0.5790290855646982,No Sick
+S_69,Male,34.0,0.4488264948240428,No Sick
+S_70,Male,52.0,0.31922894933435064,No Sick
+S_71,Male,48.0,0.2704984074393081,No Sick
+S_72,Female,57.0,0.5318201811015301,No Sick
+S_73,Female,45.0,0.915191370314399,No Sick
+S_74,Male,33.0,0.16627710406505336,No Sick
+S_75,Male,57.0,0.9744135322178493,Sick
+S_76,Male,36.0,0.3890723811007448,No Sick
+S_77,Male,50.0,0.4782644098424499,No Sick
+S_78,Female,33.0,0.5124992859909004,Sick
+S_79,Male,36.0,0.11506051247073344,No Sick
+S_80,Male,62.0,0.3673796432900976,No Sick
+S_81,Male,56.0,0.061334487122536044,No Sick
+S_82,Female,50.0,0.5792648132889187,No Sick
+S_83,Female,55.0,0.33218576134014133,No Sick
+S_84,Male,39.0,0.32085392248494365,No Sick
+S_85,Male,34.0,0.04450511997589546,No Sick
+S_86,Male,67.0,0.15133055497828424,No Sick
+S_87,Female,50.0,0.18709959213671942,No Sick
+S_88,Male,53.0,0.10073701418285581,No Sick
+S_89,Male,35.0,0.5110592056032592,Sick
+S_90,Male,38.0,0.8793802677769047,Sick
+S_91,Female,47.0,0.1247705954358479,No Sick
+S_92,Female,51.0,0.23236933325127848,No Sick
+S_93,Female,64.0,0.5361768277573149,No Sick
+S_94,Male,39.0,0.7145348867998527,Sick
+S_95,Female,49.0,0.8949969712274917,No Sick
+S_96,Female,51.0,0.4008792149067575,No Sick
+S_97,Female,51.0,0.37491428399612803,No Sick
+S_98,Female,39.0,0.3545989884164421,No Sick
+S_99,Male,51.0,0.26829780510576695,No Sick
+,,,,

README.md

@@ -0,0 +1,3 @@
+# everfortuneai_interview
+
+長佳智能AI工程師面試2023

上機考題_v5.txt

@@ -0,0 +1,21 @@
+第一題
+1. 目標：將HandBoneXRay.zip影像中的黑背景去掉，並轉正手部X光影像。當你成功開啟dicom檔影像之後，你會發現有部份影像不是整張都是手部X光影像，看起來像是X光影像丟在某個黑色背景上再翻拍成電子檔。例如
+22b63b2c78bc96711c0af031585f327874f56f18a5f42a6a7f004b8ce4a8ed26.dcm
+這個檔就有這個現象。如果看不出來，調整一下對比度(window level)。
+2. 交付項目：將處理好的影像存成同檔名的dicom檔，共十張。處理的程式碼，py檔或notebook檔都可以。
+3. 內容要求：請使用python處理，搭配pydicom模組。
+
+第二題
+讓使用者輸入一個由()AOTF 等字元構成的字串，程式運算輸出 T 或 F 或 E。
+符號與運算方式說明如下：T 代表 True，F 代表 False，A 代表 and，O 代表
+or，運算優先順序為括號內先運算、由左到右運算。
+TAT = TOF = FOT = T
+TAF = FAT = FOF = F
+若輸入的字串語法錯誤則輸出 E。例如：
+輸入 TA(TOFAT)OF，輸出 T；
+輸入 TATT，輸出 E。
+PS.禁用eval()函數
+TOFOF
+
+第三題
+附件psudo_result.csv中是某POC (proof of concept) AI測試集的結果，這個AI是拿來用在預測急診室中的某項檢傷分類疾病，該模型的輸入是胸腔X光影像。假設你是這個AI模型的開發者，請根據這個表格整理成一份不得超過兩頁的投影片報告。報告目標至少呈現AI的performance以及建議。使用MS power point或者google投影片製作皆可。