# Review — HyperDense-Net: A Hyper-Densely Connected CNN for Multi-Modal Image Segmentation

## HyperDense-Net, DenseNet Concept in 3D Network, With Multi-Modalities

--

HyperDense-Net: A Hyper-Densely Connected CNN for Multi-Modal Image Segmentation,HyperDense-Net, by École de technologie supérieure, Xidian University,2019 TMI, Over 340 Citations(Sik-Ho Tsang @ Medium)

Medical Imaging, Medical Image Analysis, Image Segmentation

**HyperDenseNet**, a**3-D fully convolutional neural network**, is proposed that extends the definition of**dense connectivity**to**multi-modal**segmentation problems.**Each imaging modality has a path**, and**dense connections**occur not only between the pairs of layers**within the same path**but**also between those across different paths, increases significantly the learning representation.**

# Outline

**HyperDense-Net Motivations****HyperDense-Net Architecture****Results**

**1. HyperDense-Net Motivations**

## 1.1. Densely-Connected Concept

- Let
be the*xl***output of the**composed of a convolution followed by a non-linear activation function:*l*-th layer by a mapping*Hl*

- A
**densely-connected network**, originated from DenseNet,**concatenates all feature outputs**in a feed-forward manner:

- where
**[. . .]**denotes a**concatenation**operation.

## 1.2. Multi-Modal Motivation

- For simplicity, consider the scenario of
**two image modalities**. - In general,
**the output of the**can then be defined as follows:*l*th layer in a stream*s*

**Shuffling and interleaving feature map elements**in a CNN was recently found to enhance the efficiency and performance, while serving as a strong regularizer, it is therefore beneficial for intermediate layers to offer a variety of**information exchange**while preserving the aforementioned deterministic functions:

- with
being a function that*πsl***permutes the feature maps given as input**. For instance, in the case of**two image modalities**, we could have:

- to have information exchange between 2 modalities as above.

**2. HyperDense-Net Architecture**

- Each
**gray**region represents a**convolutional block**. - For simplicity, it is assumed that the
**red**arrows indicate**convolution operations**only, whereas the**black**arrows represent the**direct connections**between feature maps from different layers, within and in-between the different streams.

Thus, the

inputof each convolutional block (maps before the red arrow) is theconcatenationof the outputs (maps after the red arrow) of all thepreceding layers from both paths.

## 1.3. Multi-Modal Baselines

**Single Dense Path (Left)**: An**early-fusion**strategy is followed, in which**MRI T1 and T2 are integrated at the input**of the CNN and processed**jointly along a single path.****Dual Dense Path (Middle)**: An**Late-Fusion**strategy is followed, in which each modality is processed independently in different streams and**learned features are fused before the first fully connected layer.****Early-Fusion (Right)**: An early fusion model is used, which**combines features from different streams after the first convolutional layer**.

## 1.4. Some Details

- The sub-volumes of
**size 27×27×27**are considered for**training**,**35×35×35**non-overlapping sub-volumes during**inference**. - Cross-entropy is used as cost function:

- The network was trained for
**30 epochs**, each composed of 20 subepochs. At each sub-epoch, a total of 1000 samples were randomly selected from the training images and processed in**batches of size 5**.

# 2. Results

- Dice Similarity Coefficient (DSC), Modified Hausdorff distance (MHD), are measured.

## 2.1. iSEG Challenge

HyperDenseNetobtains thebestperformance.

HyperDenseNet outperforms baselines in both cases, achieving better results than architectures with a similar number of parameters.

HyperDenseNet typically

recovers thin regions betterthan the baselines,

- The proposed network
**ranked among the top-3 methods**in**6 out of 9 metrics**, considering the results of the first and second rounds of submissions.

## 2.2. MRBrainS Challenge

Comparing the

different modality combinations,the two-modality versions ofHyperDenseNet yielded competitive performances, although there is a significant variability between the three configurations.

HyperDenseNet with three modalitiesyieldssignificantly bettersegmentations, with the highest mean DSC values for all three tissues.

HyperDenseNet ranks firstamong competing methods, obtaining the highest DSC and HD for GM and WM.

**HyperDenseNet using three modalities**can**handle thin regions better**than its two-modality versions.

## Reference

[2019 TMI] [HyperDense-Net]

HyperDense-Net: A Hyper-Densely Connected CNN for Multi-Modal Image Segmentation

## 4.2. Biomedical Image Segmentation

**2015–2019** … [HyperDense-Net] **2020** [MultiResUNet] [UNet 3+] [Dense-Gated U-Net (DGNet)] [Non-local U-Net] [SAUNet] **2021 **[Expanded U-Net]