How can I contour regions in my spatial data?

Question

I have a spatial plot with x and y coordinates and regions(differing gene expression). I am trying to identify the edges or borders of each region to contour. The best I could come up with is this plot I generated here: Spatial Plot.

I tried to use KNN to determine the cells at the edges of each region but the contour line isnt precise enough. I've used various k values and 5 seems to be the closest I can achieve with accuracy to the border. I used the spatstat.geom, ggplot and plotly packages in R. I was expecting a single contour line per region but instead I have gaps in the identified "border" cells.

my code is as follows:

library(dplyr)
library(spatstat.geom)
library(ggplot2)
library(plotly)
library(STew)

annotations <- readRDS("Manual_annotation_3639cells.rds")
data(dlpfc)

spatial_df <- as.data.frame(dlpfc$spatial)
spatial_df$cell_id <- rownames(spatial_df)
colnames(spatial_df) <- c("x_coordinate", "y_coordinate", "cell_id")

annotations_df <- data.frame(cell_id = names(annotations), 
                             region = annotations)

annotations_df$cell_id <- gsub("Manual.annotation.", "", annotations_df$cell_id)

annotations_df$cell_id <- as.character(annotations_df$cell_id)
spatial_df$cell_id <- as.character(spatial_df$cell_id)

# Merge annotations with spatial data
spatial_annotated_df <- merge(spatial_df, annotations_df, by = "cell_id", all.x = TRUE)

# Check the first few rows to ensure it looks correct
head(spatial_annotated_df)

# Convert your spatial_annotated_df to a 'ppp' object (planar point pattern)
coordinates <- spatial_annotated_df[, c("x_coordinate", "y_coordinate")]
regions <- spatial_annotated_df$region

W <- owin(xrange=range(coordinates[,1]), yrange=range(coordinates[,2])) 
ppp_data <- ppp(coordinates[,1], coordinates[,2], marks=spatial_annotated_df$region, window=W) 

# Find the nearest neighbor for each cell
nn_index <- nnwhich(ppp_data, k = 5)

# Determine if the nearest neighbor is in a different region
is_border_cell <- mapply(function(cell_idx, nn_idx) {
  ppp_data$marks[cell_idx] != ppp_data$marks[nn_idx]
}, seq_along(ppp_data$marks), nn_index)

# Add the border cell information back to the dataframe
spatial_annotated_df$border_cell <- is_border_cell

# Filter to get only the border cells
border_cells <- spatial_annotated_df[is_border_cell, ]

# View the result
head(border_cells)

# plot 
ggplot() +
  geom_point(data = spatial_annotated_df, aes(x = x_coordinate, y = y_coordinate, color = region), alpha = 0.5) +
  geom_point(data = border_cells, aes(x = x_coordinate, y = y_coordinate), color = 'black', shape = 4, size = 3) +
  scale_color_viridis_d() +
  labs(title = "Border Cells in DLPFC", x = "X Coordinate", y = "Y Coordinate") +
  theme_minimal() +
  theme(legend.title = element_text(size = 12), legend.text = element_text(size = 10)) +
  guides(color = guide_legend(title = "Region"))



# trying plot_ly to plot
fig <- plot_ly(spatial_annotated_df, x = ~x_coordinate, y = ~y_coordinate, color = ~region, type = 'scatter', mode = 'markers', marker = list(size = 10, opacity = 0.5)) %>%
  layout(title = "Border Cells in DLPFC", xaxis = list(title = "X Coordinate"), yaxis = list(title = "Y Coordinate"))