release as v2.0.0

README.md

@@ -1,5 +1,5 @@
 # MedMNIST 
-## Project Page ([Website](https://medmnist.github.io/)) | V2 Paper ([arXiv](#TODO)) | V1 Paper ([ISBI'21](https://arxiv.org/abs/2010.14925)) | Dataset ([Zenodo](https://doi.org/10.5281/zenodo.5208230))
+## Project Page ([Website](https://medmnist.github.io/)) | Dataset ([Zenodo](https://doi.org/10.5281/zenodo.5208230)) | V2 Paper ([arXiv](#TODO)) | V1 Paper ([ISBI'21](https://arxiv.org/abs/2010.14925)) 
 [Jiancheng Yang](https://jiancheng-yang.com/), Rui Shi, [Donglai Wei](https://donglaiw.github.io/), Zequan Liu, Lin Zhao, [Bilian Ke](https://scholar.google.com/citations?user=2cX5y8kAAAAJ&hl=en), [Hanspeter Pfister](https://scholar.google.com/citations?user=VWX-GMAAAAAJ&hl=en), [Bingbing Ni](https://scholar.google.com/citations?user=eUbmKwYAAAAJ)
 
 We introduce *MedMNIST v2*, a large-scale MNIST-like collection of standardized biomedical images, including 12 datasets for 2D and 6 datasets for 3D. All images are pre-processed into $28\times 28$ (2D) or $28\times 28\times 28$ (3D) with the corresponding classification labels, so that no background knowledge is required for users. Covering primary data modalities in biomedical images, MedMNIST v2 is designed to perform classification on lightweight 2D and 3D images with various data scales (from 100 to 100,000) and diverse tasks (binary/multi-class, ordinal regression and multi-label). The resulting dataset, consisting of 708,069 2D images and 10,214 3D images in total, could support numerous research / educational purposes in biomedical image analysis, computer vision and machine learning. We benchmark several baseline methods on MedMNIST v2, including 2D / 3D neural networks and open-source / commercial AutoML tools. 
@@ -27,7 +27,7 @@ Please note that this dataset is **NOT** intended for clinical use.
     * [`getting_started.ipynb`](examples/getting_started.ipynb): Explore the MedMNIST dataset with jupyter notebook. It is **ONLY** intended for a quick exploration, i.e., it does not provide full training and evaluation functionalities. 
     * [`getting_started_without_PyTorch.ipynb`](examples/getting_started_without_PyTorch.ipynb): This notebook provides snippets about how to use MedMNIST data (the `.npz` files) without PyTorch.
 * [`setup.py`](setup.py): The script to install medmnist as a module
-* [EXTERNAL] [`MedMNIST/experiments`](https://github.com/MedMNIST/experiments): training and evaluation scripts to reproduce experiments in our paper, including PyTorch, auto-sklearn, AutoKeras and Google AutoML Vision together with their weights ;)
+* [EXTERNAL] [`MedMNIST/experiments`](https://github.com/MedMNIST/experiments): training and evaluation scripts to reproduce both 2D and 3D experiments in our paper, including PyTorch, auto-sklearn, AutoKeras and Google AutoML Vision together with their weights ;)
 
 # Installation and Requirements
 Setup the required environments and install `medmnist` as a standard Python package:

examples/dataset_without_pytorch.py

@@ -1,4 +1,5 @@
 import os
+import random
 import numpy as np
 from collections import Sequence
 from PIL import Image
@@ -90,6 +91,15 @@ def download(self):
                                'Go to the homepage to download manually. ' +
                                HOMEPAGE)
 
+    @staticmethod
+    def _collate_fn(data):
+        xs = []
+        ys = []
+        for x, y in data:
+            xs.append(np.array(x))
+            ys.append(y)
+        return np.array(xs), np.array(ys)
+
 
 class MedMNIST2D(MedMNIST):
 
@@ -273,3 +283,29 @@ class SynapseMNIST3D(MedMNIST3D):
 OrganMNISTAxial = OrganAMNIST
 OrganMNISTCoronal = OrganCMNIST
 OrganMNISTSagittal = OrganSMNIST
+
+
+def get_loader(dataset, batch_size):
+    total_size = len(dataset)
+    print('Size', total_size)
+    index_generator = shuffle_iterator(range(total_size))
+    while True:
+        data = []
+        for _ in range(batch_size):
+            idx = next(index_generator)
+            data.append(dataset[idx])
+        yield dataset._collate_fn(data)
+
+
+def shuffle_iterator(iterator):
+    # iterator should have limited size
+    index = list(iterator)
+    total_size = len(index)
+    i = 0
+    random.shuffle(index)
+    while True:
+        yield index[i]
+        i += 1
+        if i >= total_size:
+            i = 0
+            random.shuffle(index)

medmnist/__init__.py

@@ -1,4 +1,4 @@
-from medmnist.info import __version__, HOMEPAGE
+from medmnist.info import __version__, HOMEPAGE, INFO
 from medmnist.dataset import (PathMNIST, ChestMNIST, DermaMNIST, OCTMNIST, PneumoniaMNIST, RetinaMNIST,
                               BreastMNIST, BloodMNIST, TissueMNIST, OrganAMNIST, OrganCMNIST, OrganSMNIST,
                               OrganMNIST3D, NoduleMNIST3D, AdrenalMNIST3D, FractureMNIST3D, VesselMNIST3D, SynapseMNIST3D)

medmnist/info.py

@@ -1,4 +1,4 @@
-__version__ = "1.3"
+__version__ = "2.0.0"
 
 
 import os