. 2017 Sep 7;12(9):e0184290. doi: 10.1371/journal.pone.0184290

Table 5. Lung nodule image sequence segmentation algorithm.

Algorithm 5 Lung nodule image sequences segmentation algorithm
1: Input the original CT image sequences, and initialize the number of slices of AIP reconstruction area S_r and the total number of AIP reconstruction sequence images N_r. 2: Lung parenchyma image sequence segmentation. 3: Obtain the AIP sequence images *AIP_i, i ∈ [1,N_r]. The lung parenchyma image sequences are labeled ${C T_{i j}}_{j = 1}^{S_{r}}$ , with i ∈ [1,N_r], and j ∈ [1,S_r]; j represents the corresponding relationships of the reconstructed sequence number. 4: For lung parenchyma images labeled {CT_i1}, with i ∈ [1,N_r], execute the following: 1) Compared with the AIP_i image sequence after multi-scale dot enhancement, retain the dot-like object. 2) Superpixel sequence image segmentation (HMSLIC). 3) Calculate the distance between superpixels and perform the Determine the Threshold* algorithm to obtain the starting clustering point, the starting clustering block and the clustering threshold, denoted as [x_i, y_i], {S_i1} and {E_i1}. 5: For lung parenchyma images labeled ${C T_{i j}}_{j = 2}^{S_{r}}$ , with i ∈ [1,N_r], and j ∈ [2,S_r], take [x_i, y_i] as the center and 30 mm as the side to extract ROIs as follows: 1) Convert coordinates [x_i, y_i] of the clustering starting point to the ROI as [X_i, Y_i]. 2) Superpixel sequence image segmentation (*HMSLIC). 3) Mark the label of the pixel point [X_i, Y_i], which belongs to the superpixel as the starting clustering block ${S_{i j}}_{j = 2}^{S_{r}}$ . 4) Obtain the clustering threshold by performing the Determine the Threshold* algorithm labeled as ${E_{i j}}_{j = 2}^{S {}_{r}}$ . 6: According to the clustering starting blocks ${S_{i j}}_{j = 1}^{S_{r}}$ and clustering thresholds ${E_{i j}}_{j = 1}^{S_{r}}$ , perform the *Improved DBSCAN* algorithm to obtain the lung nodule mask sequence ${M_{i j}}_{j = 1}^{S_{r}}$ . 7: According to ${M_{i j}}_{j = 1}^{S_{r}}$ , sequential segmentation of all lung nodule images.

Algorithm 5 Lung nodule image sequences segmentation algorithm

1: Input the original CT image sequences, and initialize the number of slices of AIP reconstruction area S_r and the total number of AIP reconstruction sequence images N_r.
2: Lung parenchyma image sequence segmentation.
3: Obtain the AIP sequence images AIP_i, i ∈ [1,N_r]. The lung parenchyma image sequences are labeled

{C T_{i j}}_{j = 1}^{S_{r}}

, with i ∈ [1,N_r], and j ∈ [1,S_r]; j represents the corresponding relationships of the reconstructed sequence number.
4: For lung parenchyma images labeled {CT_i1}, with i ∈ [1,N_r], execute the following:
1) Compared with the AIP_i image sequence after multi-scale dot enhancement, retain the dot-like object.
2) Superpixel sequence image segmentation (HMSLIC).
3) Calculate the distance between superpixels and perform the Determine the Threshold algorithm to obtain the starting clustering point, the starting clustering block and the clustering threshold, denoted as [x_i, y_i], {S_i1} and {E_i1}.
5: For lung parenchyma images labeled

{C T_{i j}}_{j = 2}^{S_{r}}

, with i ∈ [1,N_r], and j ∈ [2,S_r], take [x_i, y_i] as the center and 30 mm as the side to extract ROIs as follows:
1) Convert coordinates [x_i, y_i] of the clustering starting point to the ROI as [X_i, Y_i].
2) Superpixel sequence image segmentation (HMSLIC).
3) Mark the label of the pixel point [X_i, Y_i], which belongs to the superpixel as the starting clustering block

{S_{i j}}_{j = 2}^{S_{r}}

.
4) Obtain the clustering threshold by performing the Determine the Threshold algorithm labeled as

{E_{i j}}_{j = 2}^{S {}_{r}}

.
6: According to the clustering starting blocks

{S_{i j}}_{j = 1}^{S_{r}}

and clustering thresholds

{E_{i j}}_{j = 1}^{S_{r}}

, perform the Improved DBSCAN algorithm to obtain the lung nodule mask sequence

{M_{i j}}_{j = 1}^{S_{r}}

.
7: According to

{M_{i j}}_{j = 1}^{S_{r}}

, sequential segmentation of all lung nodule images.