Segment Anything Model (SAM) on Apple Silicon M1 and M2

SAM on Mac M1 and M2

Check if your Mac with M1/M2 chip is compatible with Metal Performance Shaders (MPS). Read link below:

Training PyTorch models on a Mac M1 and M2

Insert these two lines into code to run on Metal Performance Shaders (MPS) backend.

device = 'mps' if torch.backends.mps.is_available() else 'cpu'
sam.to(device=device)

Input image. Source SAM-Medical-Imaging

Set-up

Necessary imports and helper functions for displaying points, boxes, and masks.

Import Libraries

import numpy as np
import torch
import torchvision
import matplotlib.pyplot as plt
import cv2
import pydicom

Functions

def show_mask(mask, ax, random_color=False):
    if random_color:
        color = np.concatenate([np.random.random(3), np.array([0.6])], axis=0)
    else:
        color = np.array([255/255, 200/255, 0/255, 0.6])
    h, w = mask.shape[-2:]
    mask_image = mask.reshape(h, w, 1) * color.reshape(1, 1, -1)
    ax.imshow(mask_image)   
    
def show_box(box, ax):
    x0, y0 = box[0], box[1]
    w, h = box[2] - box[0], box[3] - box[1]
    ax.add_patch(plt.Rectangle((x0, y0), w, h, edgecolor='green', facecolor=(0,0,0,0), lw=2))    

Import a medical imaging image (dcm image)

def prepare_dicoms(dcm_file, show=False):
    dicom_file_data = pydicom.dcmread(dcm_file).pixel_array

    HOUNSFIELD_MAX = np.max(dicom_file_data)
    HOUNSFIELD_MIN = np.min(dicom_file_data)

    HOUNSFIELD_RANGE = HOUNSFIELD_MAX - HOUNSFIELD_MIN

    dicom_file_data[dicom_file_data < HOUNSFIELD_MIN] = HOUNSFIELD_MIN
    dicom_file_data[dicom_file_data > HOUNSFIELD_MAX] = HOUNSFIELD_MAX
    normalized_image = (dicom_file_data - HOUNSFIELD_MIN) / HOUNSFIELD_RANGE
    uint8_image = np.uint8(normalized_image*255)

    opencv_image = cv2.cvtColor(uint8_image, cv2.COLOR_GRAY2BGR)

    if show:
        cv2.imshow("a",opencv_image)
        cv2.waitKey()

    return…